SAE International
A Constant Radius Constant Speed Simulation 0738
Methodology-Yaw Rate Control
Cwo Gee Liang
General Motors Company
to pat Y yaw mite” comering maneuver at 2
tant speed, which in tur guarantees the vehicle
A simulation methodology is developed for the Constant, ' constant radius circular pa The simplicty of
Radius Constant Speed (CRCS) analysis to predict the we jodology proves to be more efficient, accurate and
ABSTRACT
1804138 [1] soad tet peviormance. The CRCS analysis can ‘comparing with the sophisticated dnwver-in-the-loop
ach in racing a constant radius circular path
bbe used 10 predict the vehicle steady-sate handling c
characteristics such as underter, rear comering compliance,
fd coll gradieet ete. The Vane Rate Comal methodology i
‘DAMS [7] and Hyperwork (8) are used in shis work to
Be eS a one OO) aialp ec naroet nines
Re? a? Se De
eee eee eee, YAW-RATE CONTROL
Es eae
INTRODUCTION STEER CONTROL
ee ee nae, Oe te Sian. alee
wr ac teacle nae emcee abate reac are
So gees SE aes ae
Cane dins ta ead AE ares a A ce aaa as gee neat
CEST, GotnalaadGrenrcmcnsces Si ancien oer tas dr ane
Serr ae: | Baur
Se ee ae
Ce ee oe sae pan) km, wt FL chet are
while driving the vehicle
constantadive path,
svsively higher speeds oa the ona “constant radius” czculr path if
je the test procedure, a
troditional simulatongpprtach is to incorporate 2 duver @,- Vet Roy
model (2.3. 4. 5, form the diver in-the-loop path o
fhe quality of etter rad test or
silat PEW he patcping aby ofthe ever ‘Where ais the yaw sate, V isthe tangent velocity. and Ry is
the tuning cade, The celation it true only when (1) the
ssstemnis ina steady ate and 2 the sytem is na 2D plane
Tn this work, instead of using the driver-in-the-loop path ites *
folowing, a deer salen epreach 19 proposed A
sila yaw conta algo 5 ed to eee vehicle‘Note that for a vehicle wavels at a high g maneuver wih
ogra side slip angle, Vz the vehicle tangent velocity, nat
the fore-aft speed Ve
Figure £. Circular Motion- Planar Sinematies
A simple steer PID contol is used to contol the steer
‘wheel angle 6 soto regulate the voile yaw rte as segues
‘Equation (1) fora specified constant rads Ry and 3 cen
constant vehicle tangent speed Vj As shown inthe block
diagram, Figuce_2. the steering angle command can be
defined as
=P +1S~DS)@n-@.
where the actual vehicle yaw rate, op the target ya
rate a8 defined in Equation (1), § the Laplace ana
operator, PA, and D are the gsins of proportioning, inter
and derivative terms, respective
Figure 2. Ster PID. Control
SPEED CONTROL
Im this report, 2 similar
GGeplying force to the
‘vehicle The PID
te speed PID controller
3) is sed to maintain the
approach tends 10. over.
capacsty is demanded. The abore
contol could be replaced by a more
‘Two vavitions of speed profile have been developed for the
Constant Radius Constant Speed simulation
S
1. Cosstant Radius Discrete Speed (CRDS): In this cate, the
target vebicle speed Vy isa sai fnction (a sequence of
lncreasing step functions) where atthe end peach sep he
‘ehilereachet 9a steady state
2. Constant Radive lnscesing Spe
‘nereasng speed, the get veil
Fer a slowly
be defined at
V0= Vu
@
hee Vo i nia 4. Fe dpedet
Catal, tne, i 9 y'4 mall fad cnt
sin els Set geen a om
‘dy te wher Boge al fra ttelerton sy
0.058. Note that roach tends to under predict the
performance g. soll gia) As the vehicle speed
keeps system never sees to is ue steady
sate Fi eaten the vebicle argtsposd profiles for
CRDS and
Sp
“ as,
suet ut ten tm
GYRO MODEL
The kinematics Equation (1) is te only in a orizontal OC
Y) plane Since the detailed ADAMS models are mostly bust
sn 8D, a gyrolike device is necessary to measure the vencle
response projected onto the horizontal X-Y (2D) plane
A gyro body ith 2 small and neslisibla mass is atached to
the vehicle body with the folowing constraints,
+2 hall ott connects the gyro 10 the vale body
+s kinematic constrain is applied so the gyros z ais (2) is
parallel tothe global z nis (2) and
‘4 kinematic comsiraint i used co the eyo ¥ axis (48
perpendicular ta the vehicle body's x ate
“As a remit, the guvos sevelocin, yvelecity, and san
‘elecity responses will be the same a3 the vehicle body's
responses projected tothe 2D horizontal XY) plane. Gy20's
sll and puch angles a always ze and st x. 2Aisplacements will be the same as the vehicle body's in the
slobal coordinate systera,
RESULTS
COMPARISON WITH A PATH
FOLLOWING PID STEER CONTROL
APPROACH
Aa in-house Path-Following PID Steer Controller is acquired
forthe comparison. The contolles contol the stering wheel
sagle 20 that the vehicle moves on a predefined eieulae
path, Figuses 4-5 show the responses of the pathiollowing
approach (sed lines) versus the proposed Yaw-Rate contol
approach (blue). Both are the results of Constant Radius
Increase Speed simulation with small constant forward
acceleration of 0.05 2. While the higher-bandwisth path-
fallowing approach effectively traces the vehicle on the
preseribed eirular path, is responses in yaw rote and steering
angle are much more fluctuating at the higher speed
comparing to the Yaw-Rate control approach Tae pat
following approach requues adaitional pot signal processing
to obtain the supposedly steady-state results. Note that the
‘vo approaches generate two diffeent cixcular trajectories. t
is because that che Vaw-Rate control approach does not use
‘the geographical path as the control parameter
‘Yaw Rate (degisee)
Figure 4. CRIS. song Comparison (Vane
eng oe
28
Figure $c ton Resale Comparteon (Steering
Weel Angle)
Ol 2 WITH THE DRIVER-IN-
q OP APPROACH,
ver Model [4] was acquired for the comparison. The
GPa ene san rte non
Soe arr eae ee
MS enctal ine) toss”ae tara an
8 approach (blue). It is a Constant Radius Discrete Speed
smlation in which the vehicle speed initially stays at 90 kph
for 20 sec and then increases to 100 kh for another 20 sec
‘While ILU aver models less effective in following the path
as the speed transitioning ffom $0 kph to 100 kph it
sevecthelese reaches ite steady state, Note tht st 100 kph. the
vehicle is approaching te maxioum possible Iseral
acceleration. A futher investigation of te steady state results
canbe found inthe following section
Yaw Rate.
Figure 6, CRDS Result Comparison (Yar Rate)ie
Figure 7 CRDS Result Compartson (Steering Wheel
Angle)
ACCURACY
1 this section, the accuracy is measured by how closely the
ohio stays on a 100-m ewcular path fo the steady state
Figite § shows the vehicle trajectory for a CRDS run, ia
which the vehicle speed increases fom 10 kph upto 100 kp
[epmisly than ineremont of 10 leh, Note that 28 show
in Figume § the vehicle isnot following 9 unique preset
cuealae path, rather the vebcle only comveraes to 3 1
turning radius ram at each steady fate, te. at 3 different
coc seed eile wveg os i caer
uajectery (he same 100 m rads but 2 dies
center) nits steady state
Jn ie 2 pr of tery is high eggpt Which
represents the trajectory erween 93 see an
vehicle isan a steady sate a the ey a kane
fing of the 95.95 see tajeciory sep thatthe
pair converge oa cle centered at Set
100.1907, with the radius R= 109927m, and the masta
path enor of 6.35e-4 m, The abo@eurve Siting isto St the
Values of cy pas tothe circu? lation
®
where the fing er
ms IRTTABSUCX-Xay? (Y-Yol-?D,
0
fioed a8
is the cave iting resus forall stead ate
ory segments of he CRDS ron a Fig st vies
cosa vee spends 1 so hat he eile taveng
DO Gra te prs 100 meter conan ao pth Gn
QEAR ES as
‘Table 1. Curve Flng Results for Crenlar Trajectories
“At Different Speeds
m6 ane
Sognen| Seed soon [eum [next
al sas [20] 130025] 0000! onset] 2nee04
af ae28 [an | 1o7as| sooe| onsage| 1.25604
af saa | 40 | 103456) s0042el_sooa07| eto
af enas [60 [154717] aos 1 206-0]
| soas | en | 204909] 00se6| sooor7| 4 e4eo
of esas | 70 | wr7e| o.709e! so0006| a0
zer5 [a0 [o2723] sone] smnoes| —1.736.00)
al esas | 100 | 137701 s009e] suoge7l — esse-ou
‘The above path error 18 used comparing the acausey
feenveea the JLU driverinchedeop approach aad tae Yaw-
‘Rate Contol ayproach shown ia Figures 6 and 7 The vehicle
‘eajectories (between 10 sez and 20 sec) of two approaches
sate fined separately to the ideal ecolar aguation 2). The
‘elt ave sown st Tabla 2. The VaweRate Contol sults nt
4 near-perfect 100 m crcle while JLU Driver path following
Tesult shows that the vehicle is waveling oa 299.62 mm exe
ath ie. an average steady-state path exo of -0.38mTable 2. Accuracy comparison
Nowe | ORE Mas Fnor
tere | Spear” | snem | yoo | Rin) | Ye
ip °
nese J vg | sors | s078 | ooswer | rete
ireke fon | 20.6 | 999 | sosisr | 1ase>
Figuies 2 and 10 show the cross plot of (1) Path Exzor Vs
Steenag Wheel Angle and Q) Lateral Acceleration Ve
Steering Wheel Angle dung the steady-state period of 10-20
see. Tho plate of staring angle isto ill the “enersy”
requied to maintain the steady states. Note that at an ideal
steady state, both ccs plots should convere to 2 sinale
point respectively. Figures 9 and 10 show thatthe Yaw Kate
Control approach (blue lines) gives a more “steady” state
result than the JLU Driver approach (ed)
Path Eror Vs Seeing Ange
be
ROBUSTNESS,
‘Miltiple vehicle models manning CRDS are wied to test the
robustness of the proposed methodology caps,
the vebicle speed increases ftom 20 Eph, 0 kp with
10 kph increment while the yawaagie (oalipller steers the
Vehicle to perform tie 100 m divs. cornering
maneuver The test vehicles inch car, a midsize
fer, 2 SUY, and the same STV
‘allt atthe tear The
vehicle handling characte
fs moreaied om 2199
disasbunon #8 changed
same PID serine
Contrl is applied
The ress ween Alleles comzge
te WOsaeagel ote seh aes avenge ea
Se gee ets Encore SUV-S00ng cos
sesaaalipye naire vc sow roe ae
‘rehiclefean 6 longer maintain che 100 m constant cadius
manedi@agt 100 kph.
‘lémuonal £00 kes
asically changes the
re the total vehicle mast
2689 kg and the weight,
font bias 10 46%, The
(0, and D=0.1) of Yaw Rate
‘vchisle models
Table 2. Robusmess Study
g
b
Ge oe wo Ge aw wm oN am ae a
Seng Ae
Figure 9. Cross Poi of Path Error Vs Stee
Figure 10. €Eross Plots of Lateral Acceleration Vs
‘Steering Angle
law Lawinte —lorowsecwr [was cor
ei a) [erode im > rom)
SUMMARY/CONCLUSIONS.
8 yawerate contol algoitim +8 proposed wo perform the
1S04138 Consist Radius Constam Speed (CRCS)
simulation. The simple 2D kinematic selationship between
the yaw rate, the speed, andthe tuming radius is applied. A
ID steer control is incorporated to regulate the projected 2D
yaw rate co that the vehicle i= moving on a constant radius
‘lecular trajectory. The simplicity af the methodology proves
to be more effcien, accurate and robust comparing with the
sophisticated deiverinthe-loop apeecach in waciaz 2
constant radius citcular path. The methodology has since
been implemented and deployed for future virtual vehicle
development