无人驾驶汽车综述论文.pdf_无人驾驶汽车论文 - CSDN文库

需积分: 0 468 浏览量更新于2023-05-29 评论收藏 6.77MB PDF 举报

身份认证购VIP最低享 7 折!

领优惠券(最高得80元）

资源详情

资源评论

资源推荐

Sensors

and

Actuators

A

319

(2021)

112566

Contents

lists

available

at

ScienceDirect

Sensors

and

Actuators

A:

Physical

journal

h

om

epage:

www.elsevier.com/locate/sna

Sensing

system

of

environmental

perception

technologies

for

driverless

vehicle:

A

review

of

state

of

the

art

and

challenges

Qiping

Chen

a

,

Yinfei

Xie

a

,

Shifeng

Guo

b,∗

,

Jie

Bai

c

,

Qiang

Shu

d

a

Key

Laboratory

of

Conveyance

and

Equipment

Ministry

of

Education,

East

China

Jiaotong

University,

Nanchang,

330013,

China

b

Shenzhen

Key

Laboratory

of

Smart

Sensing

and

Intelligent

Systems,

Shenzhen

Institutes

of

Advanced

Technology,

Chinese

Academy

of

Sciences,

Shenzhen,

518055,

China

c

School

of

Automotive

Studies,

Tongji

University,

Shanghai,

201804,

China

d

Shanghai

Tongyu

Automotive

Technology

Co.,

LTD.,

Shanghai,

201806,

China

a

r

t

i

c

l

e

i

n

f

o

Article

history:

Received

23

August

2020

Received

in

revised

form

8

January

2021

Accepted

12

January

2021

Available

online

15

January

2021

Keywords:

Driverless

vehicle

Environmental

perception

Selection

guidance

New

challenges

a

b

s

t

r

a

c

t

Environmental

perception

technology

is

the

guarantee

of

the

safety

of

driverless

vehicles.

At

present,

there

are

a

lot

of

researches

and

reviews

on

environmental

perception,

aiming

to

realize

unmanned

driving

while

ensuring

the

safety

of

human

life.

However,

the

technology

is

facing

new

challenges

in

the

new

era.

This

review

paper

attempts

to

systematically

summarize

environment

perception

technol-

ogy

and

discuss

the

new

challenges

currently

faced.

To

this

end,

we

ﬁrst

summarized

the

advantages,

disadvantages

and

applicable

occasions

of

several

commonly

used

sensing

methods

to

provide

a

clear

selection

guide.

The

new

challenges

faced

by

environmental

perception

technology

are

discussed

from

three

aspects:

technology,

external

environment

and

applications.

Finally,

the

article

also

points

out

the

future

development

trends

and

efforts

of

environmental

perception

technology.

©

2021

Elsevier

B.V.

All

rights

reserved.

Contents

1.

Introduction

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

2

2.

Technology

overview.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.2

2.1.

Vision

sensing.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.3

2.1.1.

Image

denoising

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

3

2.1.2.

Image

segmentation

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

3

2.1.3.

Image

recognition.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.4

2.1.4.

Image

restoration

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

4

2.1.5.

Image

compression

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

4

2.1.6.

Application

of

visual

sensing

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

4

2.1.7.

Summary

about

visual

sensing

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

5

2.2.

Laser

sensing

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

6

2.2.1.

Classiﬁcation

by

function

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

6

2.2.2.

Classiﬁcation

by

line

number

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

6

2.2.3.

Classiﬁcation

by

emission

waveform

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

6

2.2.4.

Summary

of

laser

sensor.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.6

2.3.

Ultrasonic

sensor

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

7

2.4.

Microwave

radar

sensing

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

7

2.4.1.

Interference

radar

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

7

2.4.2.

PWU-WB

radar.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.7

2.4.3.

SFCW

radar.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.7

2.4.4.

Summary

of

microwave

radar

sensing

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

7

∗

Corresponding

author.

E-mail

address:

Sf.guo@siat.ac.cn

(S.

Guo).

https://doi.org/10.1016/j.sna.2021.112566

0924-4247/©

2021

Elsevier

B.V.

All

rights

reserved.

Q.

Chen,

Y.

Xie,

S.

Guo

et

al.

Sensors

and

Actuators

A

319

(2021)

112566

2.5.

MMW

radar

sensing

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

7

2.6.

Fusion

sensing

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

9

2.7.

Performance

comparison

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

9

2.8.

Guidelines

for

selecting

sensors

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

11

2.8.1.

Sensor

selection

status

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

11

2.8.2.

Sensor

selection

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

11

3.

Challenges

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

11

3.1.

Technical

challenges

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

11

3.1.1.

Poor

stability

of

visual

sensors

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

11

3.1.2.

The

detection

distance

of

Lidar

is

limited

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

11

3.1.3.

Prone

to

omissions

of

MMW

radar

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

11

3.1.4.

Immature

fusion

sensing

technology

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

12

3.2.

External

environment

challenges

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

12

3.3.

Application

challenges

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

12

3.3.1.

Appearance

and

modeling.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.12

3.3.2.

Size

and

weight

limits

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

12

3.3.3.

Production

cost

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

12

3.3.4.

Ethical

decision

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

12

4.

Future

development

trend

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

14

4.1.

Sensor

material

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

15

4.2.

Sensor

sensitivity

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

15

4.3.

Lidar

solid

state.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.15

4.4.

Dynamic

and

static

detection

combined

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

15

4.5.

Multisensor

fusion

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

15

5.

Conclusions

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

15

Acknowledgments

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

15

References

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

15

Biography

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

18

1.

Introduction

The

driverless

vehicle

is

a

kind

of

intelligent

vehicle

that

mainly

relies

on

the

smart

driving

instrument

installed

in

the

vehicle

to

realize

the

driverless

control.

It

helps

to

eliminate

car

accidents

and

trafﬁc

jams

and

greatly

reduces

carbon

dioxide

emissions.

The

technical

model

of

the

driverless

vehicle

is

shown

in

Fig.

1(a).

It

is

generally

divided

into

four

basic

technologies,

namely

environ-

ment

perception

and

modeling,

positioning

and

map

construction,

path

planning

and

decision

making,

and

motion

control

[1].

Driver-

less

vehicle

can

reach

the

destination

safely

and

normatively

by

perceiving

the

surrounding

environment

and

planning

the

cor-

responding

driving

path

without

human

interference.

Therefore,

the

technology

of

environmental

perception

is

the

core

of

driver-

less

vehicle.

The

main

functions

of

environmental

perception

of

driverless

vehicle

are

based

on

lane

and

road

detection,

trafﬁc

sign

recognition,

vehicle

tracking,

behavior

analysis

and

scene

understanding

[2].

The

perception

system

of

driverless

vehicle

should

have

the

following

characteristics.

(1).

Accurate:

to

pro-

vide

the

vehicle

with

accurate

driving

environment

information;

(2).

Sturdy:

to

work

normally

and

not

easily

malfunctioned;

and

(3).

Real-time:

to

avoid

obstacles

in

time.

To

achieve

those

above

objectives,

a

variety

of

sensors

and

robust

fusion

algorithms

are

required.

[1].

The

ﬁeld

of

environment

perception

has

become

a

major

research

topic

in

the

world.

A

wealth

of

research

has

been

dedi-

cated

to

the

development

of

environment

perception

to

enhance

the

safety

of

driverless

driving

[3–6].

However,

there

are

still

many

technical

and

non-technical

challenges

before

the

real

driverless

driving.

A

comprehensive

literature

review

on

the

envi-

ronmental

perception

of

driverless

vehicle

was

provided

in

[1].

It

gave

the

latest

algorithms

and

modeling

methods

for

driver-

less

vehicle.

But

this

review

did

not

describe

the

non-technical

challenges

that

driverless

vehicle

may

face

in

the

future.

The

structure

of

driverless

vehicle

with

the

combination

of

naviga-

tion

and

positioning

was

introduced

in

[7],

and

predicted

the

future

technologies

for

environmental

perception.

But

this

paper

lacked

comparison

of

various

on-board

sensors

and

did

not

dis-

cuss

the

question

of

“how

to

choose”.

Through

research

on

the

above

mentioned

relevant

work,

two

major

questions

of

the

environment

perception

technology

are

still

existed

and

challeng-

ing.

Question

1:

The

classiﬁcation

and

summary

of

the

exist-

ing

sensor

technology

are

not

detailed

enough,

so

that

the

answer

to

the

question

of

how

to

select

a

sensor

is

not

speciﬁc

enough.

Question

2:

The

overview

of

the

new

challenges

faced

by

the

technology

is

not

comprehensive,

and

there

is

a

lack

of

overview

of

multiple

perspectives.

This

review

article

attempts

to

systematically

summarize

the

driverless

vehicle

environment

perception

technology,

provide

clearer

sensor

selection

guidance,

and

discuss

the

new

challenges

facing

the

technology

in

multiple

directions.

The

contributions

of

this

paper

are

as

follows:

For

Question

1,

the

several

conventional

sensing

methods

used

in

environmental

perception

technology

were

outlined.

The

advan-

tages

and

disadvantages

of

various

perception

methods,

as

well

as

their

applicable

occasions

are

summarized.

A

clearer

selection

guid-

ance

of

perception

methods

is

provided

in

Sec.

2.

For

Question

2,

the

new

challenges

and

current

unresolved

problems

faced

by

percep-

tion

technology

are

summarized

in

Sec.

3.

The

future

development

trend

and

direction

of

the

technology

are

summarized

in

Sec.

4

and

Sec.

5,

respectively.

2.

Technology

overview

The

environment

perception

technologies

of

driverless

vehicle

provide

necessary

driving

environmental

information

for

the

vehi-

cle.

The

schematic

diagram

of

the

environment

sensing

system

[8]

is

shown

in

Fig.

1(b).

It

is

composed

of

vision

sensing,

radar

sensing,

ultrasonic

sensing,

laser

sensing,

etc.

2

Q.

Chen,

Y.

Xie,

S.

Guo

et

al.

Sensors

and

Actuators

A

319

(2021)

112566

Fig.

1.

Overall

diagram

of

driverless

vehicle.

(a)

Four

basic

technologies

of

driverless

vehicle

[1].

(b)

Sensing

system

of

environmental

perception

[8].

2.1.

Vision

sensing

Vision

sensing

adopts

image

analysis

and

recognition

technol-

ogy

to

detect

the

driving

environment.

Image

processing

is

an

important

part

of

the

vision

sensing

technology.

It

uses

an

auto-

matic

machine

system

to

realize

the

understanding

of

complex

environmental

image.

Image

processing

technology

is

composed

of

ﬁve

modules:

image

denoising,

image

segmentation,

image

recog-

nition,

image

restoration

and

image

compression.

2.1.1.

Image

denoising

The

noise

in

the

images

disrupts

the

useful

information

sequence

in

the

image,

making

the

images

blur.

The

image

denois-

ing

is

to

improve

the

visual

effect

of

image

[9].

Ojha

et

al.

used

K-SVD

algorithm

to

obtain

a

suitable

dictionary

in

the

interference

data

and

to

extract

the

phase

image

content

effectively

[10].

The

block

diagram

of

the

algorithm

is

shown

in

Fig.

2.

In

addition,

there

are

many

other

noise

reduction

methods,

such

as

Block

Matching

and

3D

(BM3D)

[11],

Non-local

mean

noise

reduction

[12,13]

and

Discrete

cosine

transform

(DCT)

[14].

The

image

is

ﬁrstly

analyzed

by

unsupervised

principal

com-

ponent

analysis

(PCA)

[15].

Then,

each

small

picture

named

X

k

is

rearranged

in

a

one-dimensional

array

with

proximity,

and

an

initial

dictionary

named

D

(0)

k

is

selected.

The

K-SVD

algorithm

is

implemented

at

the

same

time

to

obtain

the

denoised

small

picture

named

∧

X

k

and

the

denoised

version

of

the

related

dictionary

named

D

k

.

Finally,

all

the

small

pieces

after

denoising

are

recombined

to

form

the

denoising

interferogram.

2.1.2.

Image

segmentation

Image

segmentation

refers

to

the

extraction

of

meaningful

features

from

the

image.

With

the

development

of

deep

learn-

ing

technology,

the

image

segmentation

effect

is

signiﬁcantly

improved

[16–18].

According

to

segmentation

principles,

the

image

segmentation

can

be

subdivided

into

the

following

three

methods:

graph

theory

(GT),

pixel

clustering

(PC)

and

depth

semantic

(DS).

2.1.2.1.

Image

degmentation

based

on

graph

theory.

Image

segmen-

tation

based

on

graph

theory

is

to

treat

the

image

as

a

weighted

graph

G

=

(V,

E).

Then,

the

optimal

cut

set

to

complete

the

image

segmentation

is

achieved.

Where,

V

is

a

set

of

nodes,

and

E

is

a

ﬁnite

set

of

edges

[19].

This

theory

is

widely

recognized

because

of

its

insensitivity

to

data

shape

and

robustness.

It

can

be

subdivided

into

NormalizedCut

[20],

GraphCut

[21]

and

GrabCut

[22].

2.1.2.2.

Image

segmentation

based

on

pixel

clustering.

Image

seg-

mentation

based

on

pixel

clustering

is

to

segment

the

feature

space

according

to

the

aggregation

of

pixels

in

the

feature

space,

and

then

map

them

back

to

the

original

image

space

to

obtain

the

seg-

mentation

result

[23].

An

interactive

image

segmentation

method

based

on

diffusion

mapping

was

presented

in

[24,25].

This

method

combined

with

K-means

clustering

algorithm

to

obtain

a

better

seg-

mentation

effect.

Besides,

there

are

other

methods

such

as

spectral

3

Q.

Chen,

Y.

Xie,

S.

Guo

et

al.

Sensors

and

Actuators

A

319

(2021)

112566

Fig.

2.

The

K-SVD

algorithm

block

diagram

[10].

clustering

[26],

simple

linear

iterative

cluster

algorithm

(SLIC)

[27],

Turbopixels

algorithm

[28],

etc.

2.1.2.3.

Image

segmentation

based

on

depth

demantic.

This

method

means

that

the

deeper

content

information

provided

by

the

image

needs

to

be

combined

to

assist

in

image

segmentation.

A

method

for

extracting

dense

feature

vectors

using

multi-scale

convolutional

network

was

introducted

in

[29],

as

shown

in

Fig.

3(a).

A

multimode

network

incorporating

digital

surface

models

(DSMS)

into

the

digi-

tal

ﬁlm

cloud

network

structure

was

proposed

in

[30],

as

shown

in

Fig.

3(b).

Moreover,

there

are

other

methods

such

as

Pyramid

Scene

Parsing

Network

[31],

Segmentation

Network

[32],

U-Net

method

[

33],

etc.

2.1.3.

Image

recognition

Image

recognition

refers

to

the

work

of

extracting

connotation

features,

classiﬁcation

and

structure

analysis

of

images

[34–37].

A

residual

convolution

recurrent

neural

network

was

used

to

solve

the

problem

of

scene

image

recognition

in

[38],

as

shown

in

Fig.

4(a).

Three

deep

learning

methods

based

on

YoLo

model

were

proposed

in

[39],

as

shown

in

Fig.

4(b–e).

The

results

showed

that

the

third

method

(Fig.

4(e))

was

more

effective

in

detecting

pedestrians

in

cloudy

weather.

It

was

superior

to

other

methods

in

accuracy

and

speed.

2.1.4.

Image

restoration

Image

restoration

is

to

keep

the

original

image

in

the

process

of

image

acquisition,

transmission,

preservation

and

presentation

[

42–44].

A

new

method

of

multi

frame

image

restoration

using

vari-

able

decibel

leaves

(VB)

was

proposed

in

[45].

This

method

uses

Bayesian

inference

to

automatically

adjust

parameters

and

ensure

image

quality.

2.1.5.

Image

compression

The

basic

goal

of

image

compression

is

to

achieve

the

best

image

quality

at

a

given

compression

ratio

[46,47].

Two

memory

image

compression

algorithms

based

on

region

of

interest

and

hybrid

image

scaling

technology

were

proposed

in

[48,49],

respectively.

They

used

CNN

and

limited

Boltzmann

machine

features

[50]

to

calculate

the

memorability

score

of

each

plaque

in

the

image,

and

generated

a

memorable

map

that

are

used

to

optimize

image

com-

pression.

2.1.6.

Application

of

visual

sensing

2.1.6.1.

Monocular

stereo

vision.

Monocular

stereo

vision

is

a

method

of

measuring

the

environment

by

taking

pictures

with

one

camera.

The

advantages

of

this

method

are

simple

structure

and

easy

camera

calibration

[154].

It

reduced

the

complexity

of

exist-

ing

multi-body

motions

based

on

tightly

coupled

tracking

between

detection

strategies

[51,52].

2.1.6.2.

Binocular

stereo

vision.

Binocular

stereo

vision

is

to

use

two

cameras

to

perceive

the

same

object

in

front

from

different

angles,

as

shown

in

Fig.

5(a).

Then

the

position

relationship

between

the

two

cameras

is

considered.

Finally,

a

three-dimensional

image

of

the

object

in

front

is

formed

[53–55].

2.1.6.3.

Multi-view

stereo

vision.

Multi-view

stereo

vision

is

to

use

multiple

cameras

to

perceive

the

same

object

in

front

from

different

angles,

as

shown

in

Fig.

5(b).

Then

the

position

relationship

among

multiple

cameras

is

considered.

Finally,

a

3D

image

of

the

object

in

front

is

formed

[56–58].

2.1.6.4.

Performance

comparison.

The

structure

of

monocular

stereo

vision

is

simple,

the

algorithm

is

mature,

and

the

amount

of

calculation

is

small.

However,

the

measurement

range

and

distance

of

the

camera

are

a

pair

of

contradictory

characteristics,

which

must

be

balanced

and

adjusted.

Binocular

stereo

vision

has

been

widely

used

in

mobile

robot

positioning

and

navigation,

obstacle

avoid-

ance

and

map

building.

However,

its

difﬁculty

is

the

matching

of

corresponding

points.

Multi-eye

stereo

vision

makes

full

use

of

the

information

of

multiple

cameras

and

reduces

the

error

matching.

It

improves

the

positioning

accuracy.

However,

its

matching

algo-

rithm

is

more

complex

and

needs

more

time.

The

performance

of

three

kinds

of

stereovision

in

visual

sensing

property

is

summa-

rized

in

Table

1.

4

剩余17页未读，继续阅读

评论0

wrjsxyf

粉丝: 0
资源: 1

会员权益专享

图片转文字

全年可省5，000元立即开通

最新资源

资源上传下载、课程学习等过程中有任何疑问或建议，欢迎提出宝贵意见哦~我们会及时处理！点击此处反馈