and Effects of Vehicle Body Motion on Handling Quality Improvement - - PowerPoint PPT Presentation

Driver Model Based Handling Quality Evaluation and Effects of Vehicle Body Motion

• n Handling Quality Improvement

with G-Vectoring Control (GVC)

Masato Abe

Kanagawa Institute of Technology Japan

Introduction and Outline of The Lecture

• 1. G-Vectoring Control (GVC) is a vehicle motion control in which the longitudinal

motion is controlled depending on the lateral motion.

• 2. Mazda has introduced passenger cars with GVC into real market and significant

effects of GVC on improving vehicle handling performance not necessarily during severe motion but in a normal vehicle motion by ordinary drivers have been confirmed.

• 3. A specific feature of the effects of GVC is that though the control gain of the

longitudinal motion is very small, a big performance improvement in lateral motion

• f the driver-vehicle system is available. Therefore, fundamental effects of GVC on

handling performance especially on a subtle influence on driver’s handling quality evaluation should be investigated satisfactorily.

• 4. Since the subjective handling quality evaluation by the ordinary drivers is not always

consistent and reliable, a driver model based evaluation method, which is more

• bjective and quantitative evaluation method, has been introduced.
• 5. The fundamental effect of GVC itself on the handling quality is experimentally

investigated using the experimental full drive-by-wire electric powered vehicle by the model based evaluation method.

• 6. As GVC controls the longitudinal acceleration depending on lateral acceleration, it

directly affects the body motion and it seems that a vehicle body motion has a significant effect on the vehicle handling quality evaluation with GVC. Therefore, in

• rder to investigate the effects of GVC on handling quality more in detail, the

experimental analysis how the vehicle body motion especially the pitch motion affects the effects of GVC on the handling quality evaluation is investigated.

Overview of G-Vectoring Control (GVC)

4

𝑯𝒚𝒅 = −𝒕𝒉𝒐 𝑯𝒛 ∙ 𝑯𝒛 𝑫𝒚𝒛 𝟐 + 𝑼𝒕 𝑯𝒛

𝑯𝒚𝒅：Longitudinal acceleration command, 𝑫𝒚𝒛：Control gain, 𝑯𝒛 ：Lateral acceleration

Longitudinal motion is controlled in coordination with the lateral motion

Control algorithm of GVC

Movement of ball-in-bowl on board with GVC draws G-G diagram during entering into and out of the curve

The name “G-Vectoring Control” comes from this change of the direction of resultant acceleration

Source： www.mazda.co.jp/dynamics/skyactive/interview/gvc/01/

Mazda has introduced passenger cars with GVC by engine torque control into real market

EVALUATION BY JOURNALISTS : STRAIGHT AHEAD DRIVING AT 80KPH

Measured data during Journalists test drive events

Driver A

• Driver continuously modulates steering wheel angle to keep straight ahead.

With GVC Without GVC With GVC Without GVC

Driver B

With GVC Without GVC With GVC Without GVC

Measured data during Journalists test drive events

EVALUATION BY JOURNALISTS : STRAIGHT AHEAD DRIVING AT 80KPH

• Driver continuously modulates steering wheel angle to keep straight ahead.

Driver D

With GVC Without GVC With GVC Without GVC

Driver C

With GVC Without GVC With GVC Without GVC

Published at JSAE's annual technical meeting

EVALUATION BY JOURNALISTS : STRAIGHT AHEAD DRIVING AT 80KPH

• Steering correction (standard deviation) reduces with most of the drivers due to GVC.
• Subjective comments by the drivers are;
• “Increased controllability in small steering operation”
• “More planted feel at straight ahead driving“
• “Look-ahead distance has naturally increased“
• GVC improves handling quality by responding to subtle steering operation.

Driver Model Based Handling Quality Evaluation Method

Subjective evaluation Objective and quantitative evaluation

 Y X L  L y y 

Objective path

s T e s h

h s h

L

 



1 ) 1 (





Vehicle

L



h



• +

+

y

y

Driver

Driver model for lane change

Simplified driver model for sudden lane change

• n straight road with constant lane width

 

s h y s y

L h L h

       1 1

3 driver handling parameters to be identified

Steering angle 𝜀ℎ Course deviation

Simple driver model Find 𝝊𝑴，𝒊，𝝊𝒊 to minimize the error here

Driver parameter identification

13

Target lateral displacement Deviation of course Steering angle 𝜀ℎ Current lateral displacement

Driver parameters 𝝊𝑴 : Response delay time [s] 𝝊𝒊 : Preview time [s] 𝒊 : Control gain[-]

Lateral displacement[m] SWA[deg] Time[s] Measured time history of vehicle motion

 

) 1 ( ) 1 (

L h h L

y y s h s         

Finding driver handling parameters

 

) 1 ( ) 1 (

L h h L

y y s h s        

dt dt dy h y y h dt d dt J

h L T h L h 2 2

] ) ( [           

 

, ) ( ,         

L h

J h J h J  

Error Finding the driver steer parameters, , to minimize J

L h

h   and ,

 

s h y s y

L h L h

       1 1

Measured real steering angle Calculated steering angle by driver model using measured vehicle lateral position

Relationships between handling quality and driver parameter

15

Target lateral displacement Deviation of course Steering angle 𝜀ℎ Current lateral displacement

Slow, relaxed behavior 𝜐𝑀: All the response delay of the driver during the lane change If 𝜐𝑀 is large Driving with a margin, easier is enough to complete lane change with ease Higher handling quality evaluation The handling quality evaluation is quantified

• bjectively by the parameter value of 𝜐𝑀.

Some Examples of Reflections of Vehicle Response Characteristics to Driver Parameter L – Handling Quality Evaluation

Airplane short period mode of longitudinal response

2 1 ) ( ) (

2 2 n n

s s G s s q       

Short period damping ratio Short period natural frequency

Pilot rating on the longitudinal response

good poor poor poor good

Relation of vehicle response parameters, n and  , to handling quality evaluation

good acceptable unacceptable

poor

2 1 1 ) ( ) ( ) (

2 2 n n

s s s T G s s     

  

   

2 1 1 ) ( ) ( ) (

2 2 n n r r

s s s T G s s r    



   

Road vehicle side-slip and yaw rate responses

Vehicle responses to front steer input are rewritten as:

(s) (s) 2 1 1 ) ( ) (

2 2 h h f n n

s s s T G s       

  

   

) ( ) ( 2 1 1 ) ( ) (

2 2

s s s s s T G s r

h h f n n r r

     



    2 1 2 1 ) (

2 * 2 * * 2 2 n n n n h f

s s s s s             

2 1 1 ) ( ) ( ) (

2 * 2 * * n n h

s s s T G s s     

  

    2 1 1 ) ( ) ( ) (

2 * 2 * * n n r r

s s s T G s s r    



    If front wheel active steer is set as follows: Variable response parameter vehicle in denominator, n* and  * is available from just above front wheel active steer only as follows:

1.0 , where ) (1 ), 1 (

2 1 2 n * 1 *

            

n

If then

 

) 1 ( 1 ) 1 ( ) 1 ( 2 1 1 2 1 1

2 2 2 2 2 1 2 2

s s s s n s

n n n n h f

                  

                                   

2 2 2 2 1 2 2 2 2 1 2 2 2 2

1 2 1 1 2 1 1 2 ) ( 2 1 2 1 1 2 1 1 s s s s n s s s s s s n

n n n n n n n n n n

                    

Direct steer part by driver Active steer part

Driving simulator with external motion control for variable response parameter vehicle

h



External motion Control System

V

f



 r

                           

2 2 2 2 1

1 2 1 1 2 1 1 ) ( s s s s n s

n n n n h

          

variable stability vehicle

Lateral acceleration Yaw rate

Frequency response of driving simulator with variable response parameter at V=80km/h

3.0m 3.0m 45[m] 2.5m 3.0m 3.0m 80[km/h]

Lane change test

Calculated motion in driving simulator and measured motions of moving base

• 15
• 10
• 5

5 10 15 1 2 3 4 5 6

• 0.5
• 0.4
• 0.3
• 0.2
• 0.1

0.1 0.2 0.3 0.4 0.5 1 2 3 4 5 6

Yaw rate (deg/s) Lateral acceleration (m/s2)

measured calculated measured calculated

70%

100% 130% 160% 190% 0.05 0.1 0.15 0.2 0.25 0.3 40% 70% 100% 130% 160%

? [-]

0.79 1.03 0.32 0.55 1.26 3.08 2.59 2.11 1.13 1.62

70% 100% 130% 160% 190% 0.05 0.1 0.15 0.2 0.25 0.3 40% 70% 100% 130% 160%

0.79 1.03 0.32 0.55 1.26 3.08 2.59 2.11 1.13 1.62

70% 100% 130% 160% 190% 0.05 0.1 0.15 0.2 0.25 0.3 40% 70% 100% 130% 160%

? [-]

0.79 1.03 0.32 0.55 1.26 3.08 2.59 2.11 1.13 1.62

Driver A Driver B Driver C Driver D

 n    n n n

Identified driver parameter L on N- plane (V=80km/h, LC=45m)

L L L L

70% 100% 130% 160% 190% 0.05 0.1 0.15

0.2

0.25 0.3

40%

70% 100% 130% 160%

0.79 1.03 0.32 0.55 1.26 3.08 2.59 2.11 1.13 1.62

70% 100% 130% 160% 190% 0.05 0.1 0.15 0.2 0.25 0.3 40% 70% 100% 130% 160%

0.79 1.03 0.32 0.55 1.26 3.08 2.59 2.11 1.13 1.62

 n

L

Averaged four driver parameters L on n- plane (V=80km/h, LC=45m)

   

 

 

           

s s M s s K l s n s K l s s r V K l K l Is s s K l K l

r r r f f r r f f r r f f

                       2 ) ( ) ( 2 2 2

2 2

δ δ

 

   

 

       

s s K s n s K s s r K l K l V mV s s K K mVs

r r f r r f f r f

     2 ) ( ) ( 2 2 2 {              δ δ ｝ Linear-2degree-of-freedom-model with the two active chassis controls

 

 

 

 

) ( ) ( 2 ) ( ) ( 2 2 1 ) ( ) ( 2 ) ( ) ( 2 2 s s r V K K l K l K mV s s K K K mVs K K n s s s r V lK l V ml Is s s lK Vs ml n lK s M

m r r r f f r m r r f r f r m f f r m f r f z

δ δ β δ δ δ δ β δ                                Solving inverse equations by M(s)/δ(s) and δr(s)/δ(s)

Model responses

Two active chassis controls for experimental vehicle(VSV) ・ DYC (direct yaw-moment control) : M

・ RWS (rear wheel active steer ) : δr

The Same Experiment Using Real Vehicle

Model response to steer input

2 2

1 2 1 1 ) ( ) ( ) ( s s s T G s s

n n m

ω ω δ β

β β δ

    

2 2

1 2 1 1 ) ( ) ( ) ( s s s T G s s r

n n r r m

ω ω δ

δ

    

 

l l AV V K l l l m G

r r r f 2 2

1 2 1   

 

 

l V AV G

r 2

1 1  



α

r f R r

lK V ml T 2 

r f r r f f

K K K l K l l m A   

2

2

2

1 2 AV mI K K V l

r f N n

 α 

) 1 ( 2 ) ( ) (

2 2 2

AV K mIK l K K I K l K l m

r f r f r r f f D

    α ζ

2

2 1 1 2 V K l l l m K ll IV T

r r f r r

 



Adjustment parameters αN αD αR

Model response

) ( ) ( , ) ( ) ( s s r s s

m m

δ δ β

Inverse Model

) ( ) ( , ) ( ) ( s s s s M

r

δ δ δ

M

r

δ

m



m

r

δ

V V δ

Feed-forward type of model following control algorithm

Actuator for

DYC

Actuator for

RWS

Speed sensor Steering angle sensor GPS Yaw rate sensor Acceleration sensor

micro auto box

Experimental VSV with active chassis controls

Lateral acceleration

Yaw rate

Frequency response of VSV at V=80km/h

3.0m 3.0m 30m 2.5m 3.0m 3.0m 80km/h

Lane change test

n ,  and driver handling parameter L

average value of four drivers average value of each driver

Blue : Driver A Red : Driver B Green : Driver C Gray : Driver D

Steering angle – torque characteristics

• 4
• 2

2 4

• 90
• 45

45 90 Steering reactive torque[Nm] Steering Angle[deg] Case 1 : Ks=0,Kf=0,FW=0,c=0

• 4
• 2

2 4

• 90
• 45

45 90 Steering reactive torque[Nm] Steering Angle[deg] Case 2 : Ks=2,Kf=0,FW=0,c=0

• 4
• 2

2 4

• 90
• 45

45 90 Steering reactive torque[Nm] Steering Angle[deg] Case 3 : Ks=2,Kf=0,FW=0,c=0.3

• 4
• 2

2 4

• 90
• 45

45 90 Steering reactive torque[Nm] Steering Angle[deg] Case 4 : Ks=0,Kf=10,FW=1,c=0

• 4
• 2

2 4

• 90
• 45

45 90 Steering reactive torque[Nm] Steering Angle[deg] Case 5 : Ks=2,Kf=10,FW=1,c=0

• 4
• 2

2 4

• 90
• 45

45 90 Steering reactive torque[Nm] Steering Angle[deg] Case 6 : Ks=2,Kf=10,FW=1,c=0.3

Pure spring + damping Pure spring Spring-friction Pure spring + spring-friction Without torque Pure spring + spring-friction +damping

Effects of steering torque characteristics

• n handling quality evaluation

DSP

Steering torque generation system (coneKt, TRW) Steering angle Reaction torque command Driving simulator Vehicle motion Data logging Steering angle Steering reaction torque Steering angle Motion & visual display Calculation of vehicle motion

Driver

（steering wheel）

Effects of steering torque on handling quality

• 4
• 2

2 4

• 90 -45

45 90 Steering torque[Nm] Steering Angle[deg]

Steering torque model

Lane change course

2.5m Lane change section : Lc Straight section Straight section 3.0m 2.5m

Vehicle speed [km/h] Lane change length : Lc [m] width [m] 60 40 2.5 100 40 55 140 55 70

60km/h , Lc=40m

τL [sec] τL [sec] τL [sec]

Driver A Driver B Driver C Driver D Driver E Driver F Driver G Driver H Driver I Driver J

0.2 0.16 0.12 0.08 0.04 0.2 0.16 0.12 0.08 0.04 0.2 0.16 0.12 0.08 0.04

y0 δ h y Driver Vehicle

s h

L

  1

s

h

  1

• +

τL

Driver A Driver B Driver C Driver D Driver E Driver F Driver G Driver H Driver I Driver J

τL [sec] τL [sec] 0.2 0.16 0.12 0.08 0.04 0.2 0.16 0.12 0.08 0.04

100km/h , Lc=55m

τL [sec] 0.2 0.16 0.12 0.08 0.04

Driver A Driver B Driver C Driver D Driver E Driver F Driver G Driver H Driver I Driver J

140km/h , Lc=70m

τL [sec] τL [sec] 0.2 0.16 0.12 0.08 0.04 0.2 0.16 0.12 0.08 0.04 τL [sec] 0.2 0.16 0.12 0.08 0.04

V=60km/h , Lc=40m V=100km/h , Lc=55m V=140km/h , Lc=70m Driver’s rating 6.0 5.0 4.0 3.0 2.0 1.0 Driver’s rating 6.0 5.0 4.0 3.0 2.0 1.0 Driver’s rating 6.0 5.0 4.0 3.0 2.0 1.0 V=100km/h , Lc=40m V=140km/h , Lc=55m Driver’s rating 6.0 5.0 4.0 3.0 2.0 1.0 Driver’s rating 6.0 5.0 4.0 3.0 2.0 1.0

Averaged driver’s subjective rating Averaged increase rate of L

Averaged increase rate of L and driver’s rating

Driver-vehicle system with driver model of steering angle control

s h

L

  1 s

h

  1

L

y y

Driver

Vehicle

 

 

s K

T

  1

h



Steer-By-Wire

s h

L

  1 s

h

  1

L

y y

Driver

Vehicle

 

 

K

T



h



Steering torque Steer-By-Wire

The interesting point is that the steering torque has the indirect effects on the driver steering angle control parameters as if the vehicle response characteristics to steering angle input changes, even though there is no change in the vehicle response. Actually it is clear that the steering torque has nothing to do with the open-loop transfer function in the above control block diagram.

Now back to GVC

Investigation into Effects of GVC

• n Handling Quality Evaluation

Using Full Drive-by-Wire Experimental EV

Experimental full drive-by-wire electric vehicle

Four wheels independent driving and braking All wheel independent steering Installation of steering system (No column)

Left and right Tie rods are not connected each other

Table 1. Basic specification of the vehicle

Vehicle mass m 709 [Kg] Yaw moment of Inertia Iz 512 [kg･m2] Wheel base l 2.050 [m] C.O.G. to front axis lf 1.012 [m] C.O.G. to rear axis lr 1.038 [m] Front tread df 1.415 [m] Rear tread dr 1.415 [m] Height of C.O.G. hs 0.417 [m]

Lane change course

43

2.5[m] 2.5[m] 18[m]

Course outline Running speed ： 40[km/h] Lane change width ： 2.5[m] Lane change margin ： 18[m] Either right or left strobe light flashed

Start of lane change section

Driver changes the lane to randomly flashed direction

44

Drivers for the test Experimental Condition Lower arm angle 0[deg] (Pitch-free) ・Without GVC ・With GVC

Repeated 15 times left and right

 

deg

 

deg

Free to pitch

Participants M/F Age Drive experience Drive frequency A1 M 22 4 years Weekly B1 M 21 2 years Daily C1 M 23 4 years Daily D1 M 23 4 years Weekly E1 M 21 3 years Daily

Evaluation for the “Driver model-based handling quality evaluation method”

Yaw rate [deg/s]

Effects of GVC on vehicle motion

45

Reduced steering angle Smooth yaw rate change

Time [s]

• Decel. force [N]

SWA [deg]

Typical GVC effects were found

 

deg

 

deg

Free to pitch

Effects of GVC on handling quality evaluation by 𝝊𝑴

46

Driver model-based handling quality evaluation by 𝝊𝑴 is useful!

Larger 𝝊𝑴 corresponds with higher handling quality

Driver-A1 Driver-B1 Driver-C1 Driver-D1 Driver-E1 Without GVC With GVC

 

deg

 

deg

Free to pitch

Supposed factors for good handling quality evaluation

47

The distinctive high quality evaluation of GVC is due to : Smooth acceleration transition of Gx and Gy (“g-g diagram”, 2D). Body motion, roll motion accompanied by pitch motion (Diagonal roll) Which is dominant? Which is essential for the better feeling ?

5

• 5
• 5

5 Gx Gx[m/s2] Gy Gy[m/s2]

Decelerate Accelerate

G-Vectoring  

y x G

G , G

Investigation how the vehicle pitch motion affects GVC improving the handling quality

49

4 wheel in-wheel motors All wheel independent steer Suspension lower arm can be adjusted for the anti-dive/lift angle

Upper arm Lower arm Running direction

Upper arm

How to change the body motion

50

Lower arm

• Xf
• Xr
• Xf・tanθf
• Xr・tanθr

θf θr Front Rear Instantaneous rotational center Running direction

Variable suspension geometry

Anti-dive force Anti-Lift force

Pitch motion restrain system

51

θf θr Front Rear Instantaneous rotational center

• Xf
• Xr

Pitch moment caused by longitudinal force (nose-dive)

Pitch motion was restrained even if the vehicle decelerates by GVC

• Xf・tanθf
• Xr・tanθr

Pitch motion restrain moment caused by vertical force Anti-dive force Anti-Lift force

Setup of the experimental vehicle

52

Case1 Case2

Suspension set up (Link alignment) Pitch-free Pitch-restrained

Without GVC With GVC Without GVC With GVC

G-linkage

No Yes No Yes

Diagonal roll (Nose-dive pitch motion)

No Yes No No

5

• 5
• 5

5 Gx Gx[m/s2] Gy Gy[m/s2]

Decelerate Accelerate

G-Vectoring

 

G , G

 

deg 30

 

deg 30

 

deg

 

deg

Influence of body pitch motion

53

Drivers for the test Experimental Condition Lower arm angle 0°&30° (Pitch-free & pitch-restrained) ・Without GVC ・With GVC

Repeated 15 times left and right

Participants M/F Age Drive experience Drive frequency

A2 M 27 9 years Daily B2 M 23 5 years Weekly C2 M 22 4 years Daily D2 M 23 3 years Weekly

 

deg 30

 

deg 30

 

deg

 

deg

Pitch free Pitch restrained

Some of them are different member

Case 1: Pitch motion free

54

GVC on GVC off Steer action becomes slow with GVC

 

deg

 

deg

• 0.3
• 0.2
• 0.1

0.1 0.2 0.3

0.5 1 1.5 2 2.5 3 3.5 4

pitch_angle [deg] Time [s]

Nose-dive Nose-up Steering start

Time[s]

Both G-linkage and Nose-dive were realized

Nose down pitching by control

Case 2: Pitch motion restrained

55

GVC on GVC off

Steering has not changed even with GVC on Pitch motion does not change so much with GVC and without

 

deg 30

 

deg 30

• 0.3
• 0.2
• 0.1

0.1 0.2 0.3 0.5 1 1.5 2 2.5 3 3.5 4

pitch_angle [deg]

Time [s]

Nose-dive Nose-up Steering start

Time[s]

Pitch motion caused by GVC was restrained

Case 1: Pitch motion free

56

GVC is effective consistently on handling quality improvement for the vehicle with pitch motion free

 

deg

 

deg

Larger 𝝊𝑴 corresponds with higher handling quality

Driver-A2 Driver-B2 Driver-C2 Driver-D2 Without GVC With GVC

57

Handling quality improvement are not consistent when the vehicle pitch motion restrained

 

deg 30

 

deg 30

Case 2: Pitch motion restrained

Larger 𝝊𝑴 corresponds with higher handling quality

Driver-A2 Driver-B2 Driver-C2 Driver-D2 Without GVC With GVC

Difference is small but effect is big!

58

• 0.3
• 0.2
• 0.1

0.1 0.2 0.3 0.5 1 1.5 2 2.5 3 3.5 4

pitch_angle [deg] Time [s] Nose-dive Nose-up Steering start

• 0.3
• 0.2
• 0.1

0.1 0.2 0.3 0.5 1 1.5 2 2.5 3 3.5 4

pitch_angle [deg] Time [s] Nose-dive Nose-up Steering start

Case 2: Pitch motion restrained Case 1: Pitch motion free

The nose-dive of GVC is very small, but indispensable for the good handling quality!

59

Effect of the vehicle pitch motion itself without longitudinal acceleration control

• n improving the handling quality

60

drivers

The longitudinal acceleration is not generated by the control but the vehicle body pitch motion is changed

Effect of the vehicle pitch motion itself without longitudinal acceleration control

• n improving the handling quality

Summary

64

1. A new vehicle motion control, GVC, is introduced. 2. Model based objective handling quality evaluation method is introduced and confirmed to be effective to evaluate such a subtle effect of GVC on handling quality for ordinary drivers. 3. The vehicle small pitch motion has a fundamentally significant influence on the result

• f the handling quality evaluation with GVC.

4. The yaw motion accompanied by the nose-dive pitch motion “diagonal-roll” during lane change makes the driver more perceptive to feel the responsive vehicle yaw motion. 5. Subtle effects of body motion on the driver’s handling quality evaluation seems very important to make ever-better cars – the vehicle fun to drive.

64

Thank you very much for your attentions