rolling stability control based on electronic stability...

EVS24Stavanger, Norway, May 13 - 16, 2009

Rolling Stability Control Based on Electronic Stability Program forIn-wheel-motor Electric Vehicle

Kiyotaka Kawashima, Toshiyuki Uchida, Yoichi HoriUniversity of Tokyo, Tokyo, Japan, [email protected]

Abstract

In this paper, a novel robust rolling stability control (RSC) based on electronic stability program (ESP) for electricvehicle (EV) is proposed. Since EVs are driven by electric motors, they have the following four remarkable advan-tages: (1) motor torque generation is quick and accurate; (2) motor torque can be estimated precisely; (3) a motorcan be attached to each wheel; and (4) motor can output negative torque as a brake actuator. These advantages en-able high performance three dimensional vehicle motion control with a distributed in-wheel-motor system. RSC isdesigned using two-degree-of-freedom control (2-DOF), which achieves tracking capability to reference value anddisturbance suppression. Generally, RSC and YSC are incompatible. Therefore, ESP, which is composed of estima-tion system(S1) and integrated vehicle motion control system(S2) is proposed. A distribution ratio of RSC and YSCis defined based on rollover index (RI) which is calculated in S1 from rolling state information. The effectiveness ofproposed methods are shown by simulation and experimental results.

Keywords:rolling stability control, electric vehicle, disturbance observer, two-degrees-of-freedom control, vehicle motion control,identification

Nomenclatureax, ay:Longitudinal and lateral accelerationayd:Lateral acceleration disturbanceayth:Critical lateral accelerationcf , cr:Front and rear tire cornering stiffnessCr:Combined roll damping coefficientd, df , dr:Tread at CG, front and rear axleFxfl, Fxfr, Fxrl, Fxrr:Tire longitudinal forcesFyfl, Fyfr, Fyrl, Fyrr:Tire lateral forcesFzfl, Fzfr, Fzrl, Fzrr:Tire normal forcesg:Gravity accelerationhc, hcr:Hight of CG and distance from CG to roll centerIr:Moment of inertia about roll axis (before wheel-lift-off)Ir2:Moment of inertia about roll axis (after wheel-lift-off)Iy:Moment of inertia about yaw axisKr:Combined roll stiffness coefficientl, lf , lr:Wheelbase and distance from CG to front and rearaxleM, Ms,Mu,:Vehicle, sprung and unsprung massN :Yaw moment by differential torqueV, Vw:Vehicle and wheel speedβ, γ:Body slip angle and yaw rateδ:Tire steering angleφ, φ:Roll angle and roll rateφth, φth:Threshold of roll angle and roll rate

1 Introduction

1.1 In-wheel-Motor Electric Vehicle’s Advan-tages and Application to Vehicle MotionControl

Electric vehicles (EVs) with distributed in-wheel-motorsystems attract global attention not only from the en-vironmental point of view, but also from the vehiclemotion control. In-wheel-motor EVs can realize highperformance vehicle motion control by utilizing advan-tages of electric motors which internal combustion en-gines do not have. The EV has the following four re-markable advantages [1]:

• Motor torque response is 10-100 times faster thaninternal combustion engine’s one. This propertyenables high performance adhesion control, skidprevention and slip control.

• Motor torque can be measured easily by observingmotor current. This property can be used for roadcondition estimation.

• Since an electric motor is compact and inexpen-sive, it can be equipped in each wheel. This fea-

EVS24 International Battery, Hybrid and Fuel Cell Electric Vehicle Symposium 1

ture realizes high performance three dimensionalvehicle motion control.

• There is no difference between acceleration anddeceleration control. This actuator advantage en-ables high performance braking control.

Slip prevention control is proposed utilizing fast torqueresponse [1]. Road condition and skid detection meth-ods are developed utilizing the advantage that torquecan be measured easily [1]. Yawing stability control,side slip angle estimation and control methods are alsoproposed by utilizing a distributed in-wheel-motor sys-tem [2] - [4].

1.2 Background and Purpose of the Research

The purpose of this paper is to propose integratedrolling and yawing stability control (RSC and YSC).Rollover stability is important for all classes of light-vehicles such as light trucks, vans, SUVs and espe-cially, for EVs which have narrow tread and high CGbecause EV is suitable for relatively small vehicle andhuman height does not change. According to the datafrom NHTSA, ratio of rollover accidents of pick ups’and vans’ crashes in 2002 was only 3% against wholeaccidents. However, nearly 33% of all deaths from pas-senger vehicle crashes are due to rollover accidents [5].Therefore, RSC is very important not only for ride qual-ity but also for safety.The RSC system has been developed by several auto-motive makers and universities [6] [7]. Rollover detec-tion systems, such as rollover index (RI) [8] and Time-to-rollover (TTR) [9] are proposed for mitigating criti-cal rolling motion.Every system controls braking force on each wheel in-dependently and suppresses sudden increase of lateralacceleration or roll angle. However, since braking forceis the average value by pulse width modulation con-trol of brake pad, brake system cannot generate precisetorque or positive torque. In the case of in-wheel-motor,both traction and braking force can be realized quicklyand precisely.In addition to actuator advantages, RSC is designedby utilizing two-degrees-of-freedom (2-DOF) controlbased on disturbance observer (DOB) [10]. For the ve-hicle motion control field, DOB is applied to vehicleyaw/pitch rate control [2] [3] and 2-DOF control is ap-plied to the electric power steering control [11]. Thereare three reasons to utilize DOB: 1)disturbance suppres-sion, 2)nominalize lateral vehicle model and 3)trackingcapability to reference value. DOB loop that suppressesthe effect of disturbance, is faster than outer loop thatachieves tracking capability. Designing DOB for trac-tion force is not applicable for ICEV, because enginetorque is not accurately known and long time delay ex-ists. Therefore, DOB is applicable only in case of EVs.The tracking capability and robustness for lateral accel-eration disturbance against such as side blast are real-ized by the proposed method.However, roll and yaw stabilities are incompatible.High rolling stability makes vehicle behavior understeer. On the other hand, high yawing stability to avoidvehicle side slip, vehicle roll stability is not guaranteed.In the next section, electronic stability program (ESP)on EV is introduced usingRI based on vehicle ge-ometry and dynamics model, which achieves integratedthree dimensional vehicle motion control.

2 Electronic Stability Program forElectric Vehicle

2.1 Introduction of Electronic Stability Pro-gram

Fig. 1 shows concept of ESP for EV. ESP consistsof two systems; vehicle/road state estimation system(S1) and integrated vehicle motion control system (S2).S1 integrates information from sensors (accelerome-ter, gyro, GPS, suspension stroke and steering anglesensors) and estimates unknown vehicle parameters(mass), vehicle state variables (yaw rate, lateral accel-eration, roll angle, roll rate and normal forces on tires)and environmental state variables [7] [12].According to the information from S1, S2 controls ve-hicle dynamics using RSC and YSC, pitching stabilitycontrol (PSC) and anti-slip control (ASC). According toRI, which is calculated by S1, a proper stability controlstrategy (YSC, RSC or mixed) is determined.RSC is based on DOB and nominal vehicle state is cal-culated by a controller. If there are errors between cal-culated and actual dynamics, it is compensated by dif-ferential torque.

Estimation

Electronic stability program on EV

S1

Vehicle parameter

Environmental state variables

Vehicle state variables

Motion control

S2

YSC A

S

CRSC

Vehicle

dynamics

Sensors

mass

yaw rate, ay, roll angle, roll rate, RI, Fz

bank angle, mu

Human

maneuver

PSC

PSC operates only before vehicle stops

Figure 1: ESP based on DOB

2.2 A Scheme of Integrated Vehicle MotionControl

Lateral acceleration is composed of vehicle side slip,yaw rate and longitudinal speed.

ay = (β + γ)V (1)

If constant vehicle speed is assumed and lateral acceler-ation is suppressed, yaw rate is also suppressed as longas differentiation of side slip is not controlled. Thisphysical constraint makes RSC and YSC incompatible.Therefore, rollover detection is necessary for integratedcontrol. In order to detect rollover, Yi proposedRI(0 < RI < 1) as a rollover detection [8]. WhenRIis high which means a vehicle is likely to roll over, theweight of RSC is set as high. On the other hand,RI issmall, which means a vehicle is not likely to roll over,the weight of YSC is set as high.Control algorithm is simple and given by followingequation. Fig. 2 shows block diagram of three dimen-sional integrated vehicle motion control.


N∗=f(RI,NRSC , NY SC , NDOB)=RI ∗NRSC + (1−RI) ∗NY SC + NDOB (2)

FV

PNay

N

Q

P P N

[P P N]n n

n

ay ay

ay

ayd

>

ayd

PMay

n

Kyfb PNγ

γ

RSC

NYSC

NDOB

ζ=f(RI,NRSC,

NYSC, NDOB)

[P P Nay ay

γγ [

[

[

NF

Kyff

Krff

Krfb

nγ

ay

V N

Figure 2: Block diagram of integrated vehicle motion control

3 Estimation System

S1 is composed of vehicle parameters, state variablesand environmental state variables estimation system. Inthis section, vehicle state variable estimation system ismainly introduced. According to the estimated statevariables,RI, a distribution ratio of RSC and YSC isdetermined.

3.1 Lateral Acceleration and Roll Angle Ob-server

Fig. 3 shows four wheel model and rolling model ofelectric vehicle.

Frl

Frr

Ffl

Ffr

l

lr

lf

V

fl

fr

rl

rr

dr

df1

2

3

4

(a)Four wheel model androtational direction

Pc

Pr

hcr hc

Center of gravity

Rotational center

Z

yCr KrIrMs

Mu

(b)Rolling model

Figure 3: Vehicle model

Vehicle motion is expressed as the following three lin-ear equations.Lateral motion:

MV (β + γ)=Fyfl + Fyfr + Fyrl + Fyrr

=−2cf (β +lfV

γ − δ)− 2cr(β − lfV

γ) (3)

Yawing motion:

Iyγ=(Fyfl + Fyrl)lf − (Fyfr + Fyrr)lr

=−2cf (β +lfV

γ − δ)lf + 2cr(β − lrV

γ)lr + N (4)

Rolling motion:

Mshcray = Irφ + Crφ + Krφ−Msghcrsinφ (5)

(φ < φwheel−lift−off )

Mshcray = Ir2φ−Msghcrsinφ + Msgd

2cosφ (6)

(φ > φwheel−lift−off )

Here, these motion equations need to be expressed asstate equations to design observer. Observer gain ma-trix, however, becomes 2× 4 matrix if whole equationsare combined. To reduce redundancy of designing gainmatrix, tire dynamics and rolling dynamics are sepa-rated. A matrix,Art connects two state equations.From eq.(3) and eq.(4), state equation is expressed as,

xt = Atxt + Btu, (7)yt = Ctxt + Dtu. (8)

It is noted that there is feedforward term in the transferfunction fromu to yt. Therefore, to eliminate feedfor-ward term and design stable observer,xt vector is de-fined using differential torque and steering angle as thefollowing equations, where,

xt =[

ay − c2δ ay − c2δ − b1N − c1δ]T

,

yt = ay,u =[

N δ]T

,

At =[

0 1−a0 −a1

],

Bt =[

b1 c1

a1b1 + b0 a1c1 + c0

],

Ct =[

1 0], Dt =

[0 c2

].

a0 =4cfcrl

2

MIyV 2− 2(cf lf − crlr)

Iy,

a1 =2M(cf l2f + crl

2r) + 2Iy(cf + cr)

MIyV,

b0 =2(cf + cr)

MIy, b1 = −2(cf lf − crlr)

MIyV,

c′0 =4cfcrl

MIy, c′1 =

4cfcrlrl

MIyV, c2 =

2cf

N,

c0 = c′0 − a0c2, c1 = c′1 − a1c2

From eq.(7), state space equation is,

xr = Arxr + Artyt, (9)yr = Crxr, (10)


where,xr =[

φ φ]T

, yr = φ,

Ar =[

0 1−Kr−Msghcr

Ir−Cr

Ir

], Art =

[0 0

MshcrIr

0

],

Cr =[

0 1].

These parameters are based on the experiment vehicle”Capacitor-COMS1” developed in our research group.The method to evaluate the values ofcf , cr are referredto the paper [13], and rolling parameters to the paper[10]It should be noted that lateral acceleration dynamics ex-pressed as eq.(8) is a linear time varying system de-pending on vehicle speed. The states are observable atvarious longitudinal speed except for a very low speed.In the following sections, for repeatability reason, ex-periment has been done under constant speed control.Observer gains are defined by pole assignment.

3.2 Rollover Index

RI is a dimensionless number which indicates a dangerof vehicle rollover. RI is defined using the followingthree vehicle rolling state variables; 1)present state ofroll angle and roll rate of the vehicle, 2)present lateralacceleration of the vehicle and 3)time-to-wheel lift.RIis expressed as eq. (11),

RI = C1

( |φ|φth + |φ|φth

φthφth

)+ C2

( |ay|ayc

),

+(1−C1−C2)( |φ|√

φ2 + φ2

), if φ(φ− k1φ) > 0,

RI = 0, else if φ(φ− k1φ) ≤ 0, (11)

where, C1,C2 and k1 are positive constants (0 <C1, C2 < 1).ayth is defined by vehicle geometry. Fig. 4 shows equi-librium lateral acceleration in rollover of a suspendedvehicle. It shows the relation between vehicle geom-etry such ash, d andKr and vehicle states such asφanday. From the static rollover analysis, critical lateralaccelerationayth which induces rollover, is defined.Phase plane analysis is conducted usingayth and rolldynamics (eq. 7). Fig. 5 shows phase plane plot underseveral initial condition (φ, φ) at critical lateral accel-eration. Consequently,φth and ˙φth are defined by theanalysis.

4 Integrated Motion Control System

4.1 Rolling Stability Control Based on Two-Degrees-of-Freedom Control

In this section, RSC based on 2-DOF control whichachieves tracking capability to reference value and dis-turbance suppression is introduced.For RSC, lateral acceleration is selected as controllingparameter because roll angle information is relativelyslow due to roll dynamics (about 100ms).

0 0.2 0.4 0.6 0.8 10

0.5

1

1.5

R o ll a n g le ( r a d)

Lat

eral

acc

eler

atio

n(m

/s/s

)

R o llo v e r th r e s h o ld ( w h e e ls lif t o ff )

ta n-12 h

d

2 hd

ayth

ay= K�

Figure 4: Equilibrium lateral acceleration in rollover of a sus-pended vehicle

−5 0 5 1 0

0

50

1 00

1 50

R o ll a n g le ( d e g )

Rol

l rat

e (d

eg/s

)

�th

�th

Figure 5: Phase plane plot of roll dynamics

4.1.1 Lateral acceleration disturbance observer

Based on fig. 6, transfer function from reference lateralaccelerationu, δ andayd to ay is expressed as the fol-lowing equation. Roll moment is applied by differentialtorqueN∗ by right and left in-wheel-motors. Referencevalue of lateral acceleration is given by steering angleand vehicle speed.

ay =PayNPn

Nay(Kff + Kfb)

1 + PayNPnNay

Kfbu +

Payδ

1 + PayNPnNay

Kfbδ

+1

1 + PayNPnNay

Kfbayd. (12)

Tracking capability and disturbance suppression aretwo important performances in dynamics system con-trol and can be controlled independently. On the otherhand, one-degree-of-freedom (1-DOF) control such asPID controller loses important information at subtract-ing actual value from reference one. In the control,there is only one way to set feedback gain as high to im-prove disturbance suppression performance, howeverthe gain makes the system unstable.Hence 2-DOF control in terms of tracking capabilityand disturbance suppression is applied to RSC. Pro-posed lateral acceleration DOB estimates external dis-turbance to the system using information;V , δ, N anday. Fig. 6 shows the block diagram of lateral accelera-tion DOB.


F Kfb*

Kff

V

PNay

N*

ayd^

[P P N]

[P P N]n n

ayd

ayay ay

ay ay

ay n

u

Figure 6: Block diagram of lateral acceleration DOB

F Kfb

Kff

V

PNay

N*

Q

v[P P N]

[P P N]n n

n

u

ay ay

ay ayay

ayd

>

ayd

Figure 7: Block diagram of 2-DOF for RSC based on DOB

Estimated lateral acceleration disturbanceayd and ay

are expressed as

ayd = ay − PnayNN∗ − Pn

ayδδ, (13)

ay = PayNN∗ + Payδδ + ayd. (14)

ayd =Pn

Nay

PNay

((PNay

PnNay

− 1)ay + (Payδ − Pn

ayδ)δ + ayd

). (15)

In eq. (15), the first and the second terms are modelingerrors and the third term is lateral disturbance. If mod-eling error is small enough,ayth is approximately equalto actual lateral acceleration disturbance.

4.1.2 Disturbance suppression and normalize ofroll model

Fig. 7 shows the proposed 2-DOF control for RSC.Estimated lateral acceleration disturbance is fedback tolateral acceleration reference multiplied by filterQ.

a∗y = v −Qayd. (16)

Filter Q is low pass filter and expressed as the followingequation [14]. In this study, the cut-off frequency is setas 63 rad/s.

Q =1 +

∑N−rk=1 ak(τs)k

1 +∑N

k=1 ak(τs)k, (17)

wherer must be equal or greater than relative order ofthe transfer function of the nominal plant. Substitutingeq. (16) to eq. (13), the following equation is defined.

ay = v + Pnayδδ + (1−Q)ayd. (18)

Disturbance, which is lower than the cut-off frequencyof Q and vehicle dynamics, is suppressed by DOB. Inaddition to the function of disturbance rejection, theplant is nearly equal to nominal model in lower fre-quency region than the cut-off frequency. Therefore theproposed RSC has the function of model following con-trol.

4.2 Yawing Stability Control

As fig. 2 shows, YSC is yaw rate control. Yaw ratereference value is defined by steering angle and longi-tudinal vehicle speed. Transfer function from yaw ratereference and steering angle is expressed as the follow-ing equation.

γ=PγNPn

Nγ(Kff + Kfb)1 + PγNPn

NγKfbu +

Pγδ

1 + PγNPnNγKfb

δ.(19)

5 Simulation Results

Three dimensional vehicle motion simulations havebeen conducted with combination software of Car-Sim7.1.1 and MATLAB R2006b/Simulink.At first, the effectiveness of RSC is verified. Lat-eral acceleration disturbance is generated by differen-tial torque for repeatability reason of experiments. Inthe simulation, lateral blast is generated at straight andcurve road driving, the proposed DOB suppresses thedisturbance effectively.To show the effectiveness of ESP, lateral accelerationresponse and trajectory at curving are compared. Itis shown that lateral acceleration is unnecessarily sup-pressed only with RSC, however, tracking capability toyaw rate reference is achieved by ESP.

5.1 Effectiveness of RSC

5.1.1 Vehicle Stability under Crosswind Distur-bance

Vehicle stability of RSC under crosswind disturbance isdemonstrated. At first, the vehicle goes straight and adriver holds steering angle (holding steering wheel as 0deg). Under 20 km/h vehicle speed control, crosswindis applied during 3-6 sec. Fig. 8 shows the simulationresults.

2 4 6 8 10-1

-0.5

0

0.5

1

Time(s)

Lat

eral

acc

eler

atio

n(m

/s/s

)

No RSC

W ith RSC

Disturbance

roll moment

(a)Lateral acceleration2 4 6 8 10

-0.2

-0.1

0

0.1

0.2

0.3

Time(s)

Yaw

rat

e (r

ad)

No RSC

W ith RSC

Disturbance

roll moment

(b)Yaw rate

Figure 8: Simulation result of RSC: Disturbance suppressionat straight road drive


When proposed RSC is activated, the proposed lateralacceleration DOB detects the lateral acceleration distur-bance and suppresses it.Then, disturbance is applied at curve road driving. Un-der 20km/h constant speed control as well, 180 deg stepsteering is applied with roll moment disturbance dur-ing 3-6 sec. Fig. 9 shows decrease of lateral accelera-tion since disturbance is rejected perfectly by differen-tial torque with RSC. The robustness of RSC is verifiedwith simulation results.

0 2 4 6 8 10-1

-0.5

0

0.5

1

1.5

2

2.5

3

Time(s)

Lat

eral

acc

eler

atio

n

No RSC

W ith RSC

Disturbance

roll moment

(a)Lateral acceleration

Disturbance

roll moment

0 2 4 6 8 100

0.2

0.4

0.6

0.8

1

Time(s)

Yaw

rat

e

No RSC

W ith RSC

(b)Yaw rate

Figure 9: Simulation result of RSC: Disturbance suppressionat curving

5.1.2 Tracking capability to reference value

In this section, tracking capability of RSC to referencevalue is verified with simulation results. Under 20km/hvehicle speed control, 180 deg sinusoidal steering is ap-plied and reference value of lateral acceleration is80%of nominal value. Fig. 10 shows that lateral accelera-tion follows reference value with RSC.

5.2 Effectiveness of ESP

Rollover experiment can not be achieved because ofsafety reason. Under 20km/h constant speed control,240 deg step steering is applied. From fig. 16, withonly RSC case, even though the danger of rollover isnot so high, lateral acceleration is strongly suppressedand trajectory of the vehicle is far off the road. On theother hand, with ESP case, the rise of lateral accelera-tion is recovered and steady state yaw rate is controlledso that it becomes close to no control case.

6 Experimental Results

6.1 Experimental setup

A novel one seater micro EV named ”CapacitorCOMS1” is developed for vehicle motion control exper-iments. The vehicle equips two in-wheel motors in therear tires, a steering sensor, an acceleration sensor andgyro sensors to detect roll and yaw motion. An uppermicro controller collects sensor information with A/Dconverters, calculates reference torques and outputs tothe inverter with DA converter. In this system, samplingtime is 1 (msec).Fig. 12 shows the vehicle control system and TABLE Ishows the specifications of the experimental vehicle.

4 6 8 10 12 14-6

-4

-2

0

2

4

6

Time(s)

Lat

eral

acc

eelr

atio

n(m

/s/s

)

No RSC

W ith RSC

RSC reference

Figure 10: Simulation result of RSC: tracking capability toreference

2 4 6 8 10 120

1

2

3

4

5

6

Time(s)

Lat

eral

acc

eler

atio

n (

ms/

s/s)

W ithout control

W ith ESP

W ith RSC only


2 4 6 8 10 12-0.01

0

0.01

0.02

0.03

0.04

Time(s)

Ro

ll a

ng

le (

rad

)

W ithout control

W ith ESP

W ith RSC only

(b)Roll angle

2 4 6 8 10 120

0.5

1

1.5

2

Time(s)

Yaw

rat

e (r

ad/s

)

W ithout control

W ith ESP

W ith RSC only

(c)Yaw rate

0 5 10 15 200

2

4

6

8

10

12

14

X (m)

Y (

m)

W ithout control

W ith ESP

W ith RSC only

(d)Trajectory

Figure 11: Simulation result of ESP: Step steering maneuver

At first, disturbance suppression performance andtracking capability to reference value are verified withexperimental results. Then, effectiveness of ESP isdemonstrated. In the experiment, since vehicle rolloverexperiment is not possible due to safety reason, step re-sponse of lateral acceleration and yaw rate are evalu-ated.

6.2 Effectiveness of RSC

6.2.1 Vehicle Stability under Crosswind Distur-bance

For repeatability reason, roll moment disturbance isgenerated by differential torque. Under 20 km/h con-stant speed control, roll moment disturbance is appliedfrom 1 sec. The disturbance is detected by DOB andcompensated by differential torque of right and left in-wheel motors. Here, the cut-off frequency of the lowpass filter is 63 rad/s.Fig. 13 shows disturbance suppression during straightroad driving. Step disturbance roll moment (equivalent


DSP

Controller

DSP

Controller

Charger

three-phase

200V

Capacitor

25F

LINUX

PC

AD

DA

counter

Torque

Monitor

keypad

DI

5channel

8channelCCN

Acc/Run/Start SW

Super

D N R

Shift SW D/R

PC/104

DI

DO

ADHL

Run/Stop

ErrReset

6channel

2channel 2channel

11channel

2channel

Phase

I,VI,V

ECU

PIC

Figure 12: Control system of experimental vehicle

Table 1: Drive train specification of experimental vehicle

MotorCategory IPMSM

Phase/Pole 3/12Rating power/Max 0.29kW/2kW

Max torque 100NmMax velocity 50km/h

InverterSwitching Hardware MOS FET

Control method PWM vector control

to 0.5m/s2 ∗ hcr) is applied around 1 sec. In the casewithout any control and only with FB control of RSC,lateral acceleration is not eliminated and vehicle trajec-tory is shifted in a wide range. On the other hand, in thecase with DOB, disturbance is suppressed and vehicletrajectory is maintained.Fig. 14 shows the experimental results of disturbancesuppression at curve road driving. Under 20 km/h con-stant speed control, 240 deg steering is applied and dis-turbance is applied at around 2.5 sec. In this case, datais normalized by maximum lateral acceleration. In thecase with RSC DOB, whole effect of disturbance is sup-pressed as no disturbance case. In the case withoutRSC, lateral acceleration decreases about 25% and ve-hicle behavior becomes unstable.

6.2.2 Tracking capability to reference value

In the previous section, since it was assured that the in-ner DOB loop is designed properly, tracking capabilityto reference value is verified with experimental results.180 deg sinusoidal steering is applied and reference lat-eral acceleration is80% of nominal value. The outerloop is designed with pole root loci method. Fig. 15

0.5 1 1.5 2 2.5-1.5

-1

-0.5

0

0.5

1

1.5

Time(s)

Lat

eral

acc

eler

atio

n(m

/s/s

)

W ithout control

W ith FB

W ith FB+DOB

Disturbance roll moment


Disturbance roll moment

0.5 1 1.5 2 2.5-0.2

-0.15

-0.1

-0.05

0

0.05

0.1

0.15

0.2

Time(s)

Yaw

rat

e(ra

d/s

)

W ithout control

W ith FB

W ith FB+DOB

(b)Yaw rate

Figure 13: Experimental result of RSC: Disturbance suppres-sion at straight road drive

0 2 4 6 8

0

0.5

1

1.5

2

Time(s)

No

rmal

ized

lat

eral

acc

eler

atio

n

No Dist with RSC

Dist with RSC

Dist without RSC

Disturbance

roll moment


Disturbance

roll moment

0 2 4 6 8

0

0.5

1

1.5

2

Time(s)

No

rmal

ized

yaw

rat

e

No Dist with RSC

Dist with RSC

Dist without RSC

(b)Yaw rate

Figure 14: Experimental result of RSC: Disturbance suppres-sion at curve road driving

shows that in the case with RSC, tracking capability toreference value is achieved.

6.3 Effectiveness of ESP

Effectiveness of ESP is demonstrated by experiments.For safety reason, rollover experiment is impossible.Therefore, experimental condition is the same as 5.2.Under 20km/h constant speed control, 180 deg stepsteering is applied.Fig. 16 shows that in the case with only RSC, lateralacceleration and yaw rate are strongly suppressed. Onthe other hand, in the case with ESP, yaw rate is recov-ered close to reference value. In addition, the rise oflateral acceleration is also recovered and stable corner-ing is achieved with ESP.

7 Conclusion

In this paper, a novel RSC based on ESP utilizing dif-ferential torque of in-wheel-motor EV is proposed. Ef-fectiveness of novel RSC designed by 2-DOF control isverified with simulation and experimental results. Thenincompatibility of RSC and YSC is described and ESPis proposed to solve the problem utilizingRI whichis calculated using estimated value of estimation sys-tem of ESP. Experimental results validates the proposedESP.

Acknowledgments

The author and the work are supported by Japan Societyfor the Promotion of Science.


6 8 10 12

6

4

2

0

2

4

6

Time(s)

Lat

eral

acc

eler

atio

n(m

/s/s

)

W ithout RSC

W ith RSC

Reference

Figure 15: Experimental result of RSC: tracking capabilityto reference

2 4 6 8 100

1

2

3

4

5

Time(s)

Lat

eral

acc

eler

atio

n(m

/s/s

)

W ithout control

W ith ESP

W ith RSC only


0 2 4 6 8 10

0

0.02

0.04

0.06

0.08

0.1

Time(s)

Ro

ll a

ng

le(r

ad)

W ithout control

W ith ESP

W ith RSC only

(b)Roll angle

0 2 4 6 8 10-0.2

0

0.2

0.4

0.6

0.8

Time(s)

Yaw

rat

e(ra

d/s

)

W ithout control

W ith ESP

W ith RSC only

(c)Yaw rate -5 0 5 10 15 20-5

0

5

10

15

20

X (m)

Y (

m)

W ithout control

W ith ESP

W ith RSC only

(d)Trajectory

Figure 16: Experimental result of ESP: Step steering maneu-ver

References

[1] Yoichi Hori, ”Future Vehicle driven by Electricity andControl-Research on Four Wheel Motored UOT ElectricMarch II”, IEEE Transaction on Industrial Electronics,Vol.51, No.5, pp.954-962, 2004.10

[2] Hiroshi Fujimoto, Akio Tsumasaka, Toshihiko Noguchi,”Vehicle Stability Control of Small Electric Vehicle onSnowy Road”, JSAE Review of Automotive Engineers, Vol.27, No. 2, pp. 279-286, 2006.04

[3] Shinsuke Satou, Hiroshi Fujimoto, ”Proposal of PitchingControl for Electric Vehicle with In-Wheel Motor”, IIC-07-81 IEE Japan, pp.65-70, 2007.03 (in Japanese).

[4] Peng He, Yoichi Hori, ”Improvement of EV Maneuverabilityand Safety by Dynamic Force Distribution with DisturbanceObserver”, WEVA-Journal, Vol.1, pp.258-263, 2007.05

[5] National highway traffic safety administration, Safercar pro-gram, http://www.nhtsa.gov/

[6] E. K. Liebemann, ”Safety and Performance Enhancement:The Bosch Electronic Stability Control(ESP)”, SAE Techni-cal Paper Series, 2004-21-0060, 2004.10

[7] Hongtei E. Tseng, et al, ”Estimation of land vehicle rolland pitch angles”, Vehicle System Dynamics, Vol.45, No.5,pp.433-443, 2007.05

[8] Kyongsu Yi, et al, ”Unified Chassis Control for Rollover Pre-vention, Maneuverability and Lateral Stability”, AVEC2008,pp.708-713, 2008.10

[9] Bo-Chiuan Chen, Huei Peng,”Differential-Braking-BasedRollover Prevention for Sport Utility Vehicles with Human-in-the-loop Evaluations”, Vehicle System Dynamics, Vol.36,No.4-5, pp359-389, 2001.

[10] Kiyotaka Kawashima, Toshiyuki Uchida, Yoichi Hori,”Rolling Stability Control of In-wheel Electric Vehicle Basedon Two-Degree-of-Freedom Control”, The 10th InternationalWorkshop on Advanced Motion Control, pp. 751-756, TrentoItaly, 2008.03

[11] Bilin Aksun Guvenc, Tilman Bunte, Dirk Odenthal and Lev-ent Guvenc, ”Robust Two Degree-of-Freedom Vehicle Steer-ing Controller Design”, IEEE Transaction on Control Sys-tems Technology, Vol. 12, No. 4, pp.627-636, 2004.07

[12] A. Hac, et. al, ”Detection of Vehicle Rollover”, SAE Techni-cal Paper Series, 2004-01-1757, SAE World Congress, 2004.

[13] N. Takahashi, et. al, ”Consideration on Yaw Rate Control forElectric Vehicle Based on Cornering Stiffness and Body SlipAngle Estimation”, IEE Japan, IIC-06-04, pp.17-22, 2006

[14] Takaji Umeno, Yoichi Hori, ”Robust Speed Control of DCServomotors Using Modern Two Degrees-of-Freedom Con-troller Design”, IEEE Transaction Industrial Electronics,Vol.38, No. 5, pp.363-368, 1991.10

AuthorsHe received B.C. and M.S. degrees

in E.E. department, the University of

Tokyo in 2004 and 2006, and pro-

ceeded to Ph.D. course, the University

of Tokyo. He is now researching the

motion control of electric vehicle.

He is working as a engineering official

in E.E. department, the University of

Tokyo

He received Ph.D degree in E.E. from

the UT in 1983 and became a Profes-

sor in 2000. IEEE Fellow. He is now

an AdCom member of IEEE-IES. He is

the President of IEE-Japan IAS in 08’-

09’. His research fields are control the-

ory and its industrial application to EV,

mechatronics, robotics, etc.


rolling stability control based on electronic stability...

Documents