advanced driver assistance systems (adas) development … · advanced driver assistance systems...

Research & Advanced Engineering

Advanced Driver Assistance Systems(ADAS)

-Development at Ford Motor Company

D. Gunia, T. Wey, J. MeierFord Research and Advanced Engineering


Content

� Driver Assistance and Active Safety

� Motivation

� Safety Model

� Attribute

� Technology Overview

� Customer Functions

� Sensors and Actuators

� Design Requirements

� Assessment of Driver Assistance / Active Safety

� Control Concept Development

� Example „Lane Keeping Aid“


Motivation for ADAS (= Driver Assistance and Active Safety) Technologies

Driver Assistance Technologies

• Customer demand

• Comfort improvement

• Support driver with day-to-day standard tasks

Active Safety Technologies

• Improve real-world safety

• OEM corporate citizenship / responsibility

• Safety rating programs

• Legislation

Most ADAS Systems have a comfort and an active safety portion,e.g. ACC contains cruise control and forward alert

Motivation


Historical Trends in Total Traffic AccidentFuture Product Design

0

10,000

20,000

30,000

40,000

50,000

60,000

1970 1980 1990

Year

Fata

liti

es

2000

U.S.

Japan

U.K., Germany and France

2020

FutureIncrease Expected*• Population increase• Aging Population • Fleet Shift

* Excludes effect of counter-measures

Converse trend

Need for advancedsafety technologies


What is Active Safety?Safety model

Phase 4

InCrash

Phase 2

DangerPhase

Phase 1

NormalDriving

PRIMARYActive Safety

SECONDARYPassive Safety

TERTIARYTertiary

Phase 3

CrashUnavoidable

Phase 5

PostCrash

Collision Avoidance

Occupant Protection

Rescue

IMPACT

Phase 4

InCrash

Phase 2

DangerPhase

Phase 1

NormalDriving

PRIMARYActive Safety

SECONDARYPassive Safety

TERTIARYTertiary

Phase 3

CrashUnavoidable

Phase 5

PostCrash

Collision Avoidance

Occupant Protection

Rescue

IMPACT

“Active” or “Primary” Safety

=

Accident avoidance and/or crash severity reduction

Active safety contains the 3 phases of collision avoidance

• Assistence- (phase 1),

• Warn- (phase 2) and

• Intervene-technologies (phase 2 and phase 3)

can account for collision avoidance

Active safety contains the 3 phases of collision avoidance

• Assistence- (phase 1),

• Warn- (phase 2) and

• Intervene-technologies (phase 2 and phase 3)

can account for collision avoidance


Active Safety as AttributeA

ttri

bute

Le

ve

l 3

Att

rib

ute

Le

ve

ls 4

& 5

Preventive

To Be Observed

Collision

Avoidance

Dynamic

Avoidance

To See, Safety Visibility

Forward vision

Base Braking

Safety Braking(driver apply)

Transitional

Stability

Lateral Support

Forward Threat

Base Handling

Straight Ahead

Stability

Cornering Stability

Driver Information

Driver Controls

Front

Rear Threat

Driver Status

Low SpeedDamage

To be in command

(hands on wheel, eyes onroad, mind on traffic)

Rear-end

Frontal

Intersection

Pedestrian

Rear collision

Automatic parking

Damage avoidanceby intervention

Side Threat Lane change

Road dep./rolloveraccident

Side

Rear

Side vision

Rear vision

Base Steering

Longitudinal SupportActive Driver tactical Support

Traffic / Road Info

AutonomousLongitudinal Control

AutonomousLateral Control

Active Driver Operational Support


Content


� Motivation

� Safety Model

� Attribute









“Baseline” Driver Assistance TechnologiesFord Focus / 2006

Ergonomics

selfselfselfself----dimmingdimmingdimmingdimmingrear view mirrorrear view mirrorrear view mirrorrear view mirror

rain sensing wipersrain sensing wipersrain sensing wipersrain sensing wipers

stable stable stable stable chassis setchassis setchassis setchassis set----upupupup

auto auto auto auto lightinglightinglightinglighting

parkingparkingparkingparkingsensorsensorsensorsensor

runrunrunrun----flat tiresflat tiresflat tiresflat tires

XenonXenonXenonXenonheadlampsheadlampsheadlampsheadlamps frost warningfrost warningfrost warningfrost warning

heated windscreenheated windscreenheated windscreenheated windscreen

ESPESPESPESP

hands free phonehands free phonehands free phonehands free phone

adaptive adaptive adaptive adaptive front lightingfront lightingfront lightingfront lighting

ABSABSABSABS


Functional Description

• Driver selects speed and distance level to preceding vehicle

• A sensor will spot preceding vehicles and classify them

• Vehicle automatically adjusts speed and distance respectively time gap by

means of the engine and braking systems control

• Forward Alert and Collision Mitigation functionality is included

Adaptive Cruise ControlACC



ESP improves actively the driving stability

• Acts in all driving maneuvers

• Supports driver in steering maneuvers

• Reduces yaw movement of vehicles

ESP Technical Approach

• Detection of actual driving state

• Detection of driver intention

• Reduction of difference between actual

driving state and driver intention by brake

interventions at single wheels and engine

intervention

Electronic Stability Program

T∆

�

�

�

�

T∆

�

�

�

�


Roll Over Mitigation (Part of ESP)


• Reduce the risk of untripped rollover induced

by severe steering maneuvers

• Brake front axle in case of roll over

situation to induce understeer tendency

• Reduce the risk of tripped rollover by

improving vehicle tracking stability


Semi Automatic Parallel Parking


• During parallel parking the SAPP

system does steer the vehicle

automatically. The driver just needs

to accelerate and brake the vehicle.

• Supports also multiple move

manoeuvres so that small park slots

can be targeted.

• The Park Slot is measured by

Ultrasonic Sensors.

• Only two extra sensors are required

to measure the park slot.



• The lane of the vehicle will be spotted

by a camera

• Depending on the system a single lane

marker will be sufficient to determine

the lane

• Problem: Perspective depth

cannot be recognized (2D) with

only one camera: bumpy lane

• In case of driver departing the lane

unintentionally, a warning will be

generated: audio / haptic / vision

• System in function from a speed level

of approx. 70kph and above (subject

to tuning)

Lane Departure Warning


Comfort Lane Keeping Aid

Functional Goal• Increase steering comfort on long distance

driving• Reduce number of unintended lane departures

Functional Description• Continuous and Minimal Steering Torque

Assist during normal driving• Vehicle is kept in lane intuitively at low

steering effort• Useful only on roads with limited

curvature and stable lane marking condition

Required System Developments• Camera with Special Requirements

- Lateral Offset in Lane- Yaw Angle to Lane- Curvature of Road ahead

• Simple but sound Control strategy, including camera signal filtering

• New Steering Torque Overlay Controls• HMI

Continuous Continuous Continuous Continuous Steering SupportSteering SupportSteering SupportSteering Support

Key Issues• Customer Acceptance including HMI• Driver in the Loop• How to determine Driver Wish• Functional Safety


Degree of Driver Assistance

ManualManualManualManual AssistedAssistedAssistedAssistedSemi Semi Semi Semi

AutomaticAutomaticAutomaticAutomatic

AutoAutoAutoAuto��

nomousnomousnomousnomous

Fully Fully Fully Fully

AssistedAssistedAssistedAssisted

?Semi Semi Semi Semi

Automatic Automatic Automatic Automatic

ParkingParkingParkingParking

Lane Lane Lane Lane

Departure Departure Departure Departure

WarningWarningWarningWarning

Adaptive Adaptive Adaptive Adaptive

Cruise Cruise Cruise Cruise

ControlControlControlControl

ESPESPESPESP

Lane Lane Lane Lane

Keeping Keeping Keeping Keeping

AidAidAidAid


Content


� Motivation

� Safety Model

� Attribute









How to Measure ADAS Technologies?

ADAS Attributes

• Five main classes of attributes are addressed:

• Improvements can be judged by reference to real-world accidents -verification by accident data analysis

• Objective design verification of ADAS by means of test procedures is preferred, but typically the driver is “in the loop”

• Driver acceptance is key → customer clinics including measurements of driver behavior and psychological reaction

Lighting Visibility Braking Handling Ergonomics


ESP Objective Testing

Objective evaluation in CAE and in-vehicle

• Example “NHTSA avoidance

maneuver”: single sine steer with

increasing amplitude (with dwell)

• Open-loop: steering robot

• Assessment of agility and stability

in one manoeuvre

• Sideslip angle / yaw rate and

lateral acceleration as judgement

criteria

Testing & Validation

CAEIn-

Vehicle


Semi Automated Parallel Parking

Objective evaluation in CAE

• Parking scenario definition

• Variation of typical parameters:

e. g. start parking position

• Parking result (distance to curb,

angle vs. curb) as judgement

criteria

Testing & Validation

CAE


VIRTTEX – Driving SimulatorAssessment of driver in the loop

• 24’ Dome

• Hydraulic motion system

• Simulates non-emergency driving

• 180°forward FOV

• 120°rear FOV

• Covers all mirrors


Content


� Motivation

� Safety Model

� Attribute








Continuous Continuous Continuous Continuous Steering SupportSteering SupportSteering SupportSteering Support



Control Overview

Vehicle Position in

LaneVehicle

++++ ++++

Driver

Steering System TorqueFeedback (+Noise)

++++

Steering Torque

Delta Steering Angle Request

Camera

cLKAAngle

Controller

++++

++++

Steering Torque Overlay

Driver Torque

cLKATorque

Controller

++++

Driver Wish

Estimator

��

Steering Angle

Steering

++++��



Steering Angle Controller

Average

Radius rC

of Camera

Road Shape

Relative

Yaw Angle

dΦ

Vehicle

Speed v

Road

Shape

Lateral

Displacement

dY

Reference

Steering Angle

δR

Look Ahead

Distance D

Effective Distance of Camera

DC

Camera

Road

Shape

= Look Ahead Point

Positioned on Camera

Road Shape at Look Ahead

Distance in front of Camera

C

YR

rWBd

D

d 1⋅++= φδ

Wheel Base

WB

Camera Signal



Angle Controller Overview

Vehicle++++

Driver

Driver Wish

cLKATorque

Controller

Camera

��

Steering

dy (relative)

dΦ (relative)1/rCCamera

Performance:

- D

- Driver Wish Evaluator

KC

KY = 1/D

KΦ

= 1

KC = WB

KΦ

KY


0 5 10 15 20 25 30 35 40 45 50-2

0

2

4

6

8

10

0 5 10 15 20 25 30 35 40 45 50-20

-15

-10

-5

0

5

10

15


Quality of Camera Signals

Yaw Rate

Snake Manoeuvre on straight Road

dY

plot (time(1:50000), yawrate(1:50000), 'DisplayName','YawRate', 'YDataSource', 'YawRate'); figure(gcf)

Time [s]

dy [m]

Yaw Rate[deg/s]


0 5 10 15 20 25 30 35 40 45 50-20

-15

-10

-5

0

5

10

15


Quality of Camera Signals

Yaw Rate

Snake Manoeuvre on straight Road

dY

0 5 10 15 20 25 30 35 40 45 50-0.2

0

0.2

0.4

0.6

0.8

1

1.2

plot (time(1:50000), yawrate(1:50000), 'DisplayName','YawRate', 'YDataSource', 'YawRate'); figure(gcf)

dΦ

Time [s]

dΦ [rad]

Yaw Rate[deg/s]



Filter Concept

Vehicle Speed

Curvature

FilterCurvature Raw

Yaw Rate

dy Raw

dPhiEstimator

dPhi Raw

Vehicle Speeddy

Estimator

To Angle

Controller

To Angle

Controller

To Angle

Controller

Camera Signal

ESC Signal

!!!!



Luenberger Observer

u(t)System

C

+

x(t)

B

K

A

∫∫∫∫

x(t)^-

q(t)^

Estimator

q(t)^.



dY Estimator used for cLKA

φdvdY ⋅=&

q(k) = q(k-1) . T + q(k-1)

q(k) = K . R + v(k) . dΦ(k) , with K=1

R = R ( v(k), dΦ(k), dY(k)- dY(k) )

u

+

x

x̂

-

q̂

Estimator

1

ResidualResidual

Checker

1

C

K

=

)(

)()(

td

tvtu

φ

)()( tdtx Y=

0

A

∫∫∫∫

^ ^.

q̂

.

^

^.

^

from ESC

from Φ-Estimator

Relative Yaw Angle

dΦ

Vehicle

Speed v

Lateral Displacement

dY

Relative Yaw Angle

dΦ

Vehicle

Speed v

Lateral Displacement

dY

)(ˆ)(ˆ tdtq Y=

System


1

dyEstimated

1

0.1s+1

Transfer Fcn1

1

0.15s+1

[vy]

[dyEst]

[StepTime]

[dyEst]

[dyEst]

[vy]

[vy]

[StepTime]

[StepTime]

D-Time

0

0

0

|u|

3

Psi_Est[rad]

2

VehSpeed[m/s]

1

dy_Raw[m]


Residual Checker used for cLKA

dY (k) = dY (k-1) + vY (k) + R(k) . T

R(k) = sign(Dif) . min( |Dif|, Lmax ), Lmax= const.

Dif(k) = ( dY (k) - dY (k-1) ) / T - vY(k)

Residual Checker

^


0 5 10 15 20 25 30 35 40 45 50-1.5

-1

-0.5

0

0.5

1

1.5


dY Estimator

Snake Manoeuvre on Straight Road

dY Estimated

dY

Time [s]

dy [m]



dΦ

Estimator used for cLKA

x

-

q̂

1

=

)(

)(

)(

)(

tv

tcurv

t

tu

ϕ&

)()( tdtx φ=

from ESC

1/rC from Camera

from ESC

ϕφ && +⋅= vcurvd

Average

Radius rCof Camera

Road Shape

Relative Yaw Angle

dΦ

Vehicle

Speed v

CameraRoad

Shape

Vehicle

Yaw

Rateφ.

Average

Radius rCof Camera

Road Shape

Relative Yaw Angle

dΦ

Vehicle

Speed v

CameraRoad

Shape

Vehicle

Yaw

Rateφ.

u

+

x̂

Estimator

ResidualResidual

Checker

1

C

K

0

A

∫∫∫∫q̂

.

)(ˆ)(ˆ tdtq φ=

System


0 5 10 15 20 25 30 35 40 45 50-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2


dΦ

Estimator

Snake Manoeuvre on Straight Road

dΦ Estimated

dΦ

Time [s]

dΦ [rad]



Conclusions

• Simple Controller with feed forward delivers good control

results.

• Sensor Signal Filtering is key.

• Safe Environment Sensing and Modelling is a challenge.

• To take the driver in the Loop is a challenge.

• Complex Sensor and Actuator Technologies are getting more

and more into mass production.

• Driver Assistance will change the way of driving dramatically

and soon.

• How can Control Theory help Driver Assistance in future ?


Thank You !Thank You !Thank You !Thank You !

[email protected]@[email protected]@ford.com

Back Up Back Up Back Up Back Up …………....


Radius of Path r

Yaw Angle

φVehicle Speed v

Steering

Angle

δ

Wheel Base

WB

rWB

1)sin( ≈≈δδ

rc

1=

cWB ⋅≈δ

Steering

Angleδ


Relative

Yaw Angle

dΦ

Vehicle

Speed v

Lateral

Displacemen

t dY


Camera Signal

Average

Radius rC

of Camera

Road Shape

Relative

Yaw Angle

dΦ

Vehicle

Speed v

Camera

Road

Shape

Vehicle

Yaw

Rateφ.

advanced driver assistance systems (adas) development … · advanced driver assistance systems...

Documents