Professional Documents
Culture Documents
Art. Burgio
Art. Burgio
Art. Burgio
1. Introduction
The vehicle motion control problem has been broadly
addressed in the literature and established solutions
are already in the market.
In the brakes stand-alone case, the most common
control approach is linear Proportional and
Proportional Derivative controller (P/PD) like with
gain-scheduling, which guarantees simplicity of design,
aordable in vehicle tuning and robustness, but the
extension of these controllers for the integrated case is
dicult due to their local validity in the neighbourhood
of instability points (van Zanten et al. 1995, 1998). Brake
stand-alone vehicle controllers are at the moment the
most robust solution for improving vehicle stability.
The status for steering stand-alone vehicle controllers
is less consolidated.
Market available vehicle steering controls are feed
forward like variable gear ratio (where the steering
ratio is changed to higher values for higher vehicle
speed) and -split braking (compensate with the steering
the yawing coming from the dierent leftright braking
force) are examples, but the car manufacturers are
clearly interested in steering feedback controllers to
improve handling and stability (Koehn and Eckrich
1998). Plenty of solutions are instead proposed in the
literature, like yaw control with yaw rate-lateral
velocity decoupling (Marino and Conili 2004), or with
***
2005 style
[21.12.20051:58pm]
[Invalid folder]
[162169]
F:/T&F UK/Tcon/TCON_I_79_02/TCON_A_148880.3d
(TCON)
TCON_A_148880
163
kij
Fzij
D
C
MB
>
>
Fyr x, Fzr: , kr: Fyr r , Fzr: , kr:
>
>
>
>
>
>
vy l f v
>
>
>
,
F
,
k
F
yrl
zrl rl
>
>
vx
>
>
>
>
>
>
vy lf v
>
>
:
Fyrr
, Fzrr , krr :
vx
3. Controller structure
Rewriting the plant equations in state space form, the
control problem can be stated as
CP: given the plant
v_y vx v
v_
m
J
l f, l r
vx
vy
v
x [vy, v ]0
***
2005 style
[21.12.20051:58pm]
[Invalid folder]
[162169]
1
where
9
Fyf x,D C , Fzf: ,kf: Fyr x,Fzr: ,kr: >
>
=
m
MB >
>
;
Fyf x,D ,C ,Fzf: , kf: lf Fyr x, Fzr: ,kr: lr
J
J
3
dv
| < tolerance
F:/T&F UK/Tcon/TCON_I_79_02/TCON_A_148880.3d
(TCON)
TCON_A_148880
164
If
is D invertible for every feasible value of yaw rate, lateral velocity, longitudinal velocity, vertical forces (left
and right) and longitudinal slips (left and right);
hyp.2: zero dynamics stability
dened the new state variable
vyNS vy lNS v ,
lNS
J
mlf
Driver
SWA
Limit targets
DE
Targets
VC
CT
Measurements
Vehicle
MzBrakes
VSO
Estimations
Plant
Controller
Figure 1.
***
2005 style
[21.12.20051:58pm]
[Invalid folder]
[162169]
F:/T&F UK/Tcon/TCON_I_79_02/TCON_A_148880.3d
(TCON)
TCON_A_148880
165
F
>
yNS
yr
>
lf
vx
<
>
>
>
>
:
mvx v
a typical case
, Fzr: , kr:
vyNS 0 0
6
vy l f v
1
J
Fyf1
lr Fyr r , : v
lf
vx
v v_
v
v
7
8
lr
1
Fyr
Jv
lf
lf
which, under Hip.1, gives equations (7) and (8) for the
steering control variable C and guarantees asymptotic
stability for the yaw rate tracking error.
Moreover, using the new dened state variable vyNS,
the lateral velocity equation in (3) is substituted by
mv_yNS
m 1961;
J 3700;
. lf 1.36;
. lr 1.54;
.
.
mvx v :
lNS
9
Figure 2.
***
2005 style
[21.12.20051:58pm]
COG
PNS
vyNS lNS lr v
lf lr
Fyr
, Fzr: , kr:
lf
vx
[Invalid folder]
[162169]
v
vy dynamics
vyNS (zero)
dymamics
Figure 3.
vy
vyNS
F:/T&F UK/Tcon/TCON_I_79_02/TCON_A_148880.3d
(TCON)
TCON_A_148880
166
2. Hypothesis 2 is valid if the rear tyres are not saturating, as Fyr is a monotonically increasing function of
vyNS, but if saturation occurs and lFyr <
lf mvx v , then the state vyNS integrally diverges.
This shows a physical limit of the system, independent from the adopted steering control. From the
vehicle dynamics point of view, it means that the
equilibrium of the state vyNS is obtained only when
the force generated by the rear tyres balances the centrifugal force. But the centrifugal force grows with
the yaw rate and the vehicle longitudinal speed, is
unbounded, while tyre force is limited by the tyre
characteristic; therefore if the yaw rate or the longitudinal speed are too high, the lateral velocity in the
point PNS will integral like diverge. This event
needs to be avoided either limiting the longitudinal
speed or limiting the target yaw rate, otherwise the
vehicle will spin.
3. The available friction coecient is used in the feedback loop to divide the target yaw moment, therefore
resulting in higher feedback gain for low values.
This is not in line with the fact that, in slippery
surfaces, the experienced driver gives smoother
steering corrections. Therefore the above division
is avoided, resulting then in a lower controller
bandwidth for low .
Both hypotheses for applicability of SFL steering controller are related to the saturation of front or rear
tyre, in particular the saturation of the front tyre is
related to the performance of the yaw tracking (loss of
controllability) while the saturation of the rear tyre is
related to stability. These are properties well known to
vehicle dynamics experts.
The active steering control law c(x) is chosen according to (7), (8) in all the cases the steering actuator can
provide good yaw rate tracking and vehicle stability,
i.e. when the front and rear tyres are not saturating.
This is in line with the previously stated steering
priority assumption.
. If the front tyres saturate and the yaw control system
is still asking for higher lateral forces, then the braking
.
2005 style
[21.12.20051:58pm]
[Invalid folder]
[162169]
OV
ENDIF
where KPvy > 0 is a tuning parameter regulating a
proportional term on vyNS.
The previously proposed steering controller is then
modied to include the brake stability correction in
the following way:
v v_
FyfT
v
v
lr
1
Fyr r , Fzr: , kr:
Jv MB
lf
lf
C D
The multiple output control approach extends the proposed SISO design with the objective of obtaining a
full state control which properly integrates the brakes
action on the dened steering controller, still satisfying
the assumptions A1, A2.
A natural choice for the integrated controller is
that the additional brakes actuator action in
u(x) [c(x), MB(x)] is chosen to support the cases
when the applicability hypothesis 1, 2 are not valid.
***
10b
OV
10c
vy l f v
Fyf1 FyfT , Fzf: , kf:
vx
10d
UN
ENDIF
F:/T&F UK/Tcon/TCON_I_79_02/TCON_A_148880.3d
(TCON)
TCON_A_148880
167
UN
MB
10f
OV
Fyf
MB
J
4 lf
0
2
31
30
lr
m
Fyr
B v1
6
7C
4 l J
5@
5A
v2
F
v
v
mlf
yr
x
mlf
12
13
The same controller can be found applying MIMO feedback linearization design method to the vehicle model
and slightly modifying the brakes control law, according
to assumption A2.
The vehicle model (3) is written in the form
x_ f x gxu
with
"
"
;
vyNS
2
Ffy
lr
Fyr
J
6
6
fx 6
4 l
Fyr vx v
mlf
MB
3
7
7
7;
5
lf
6 J
6
gx 6
4
0
1
J
9
>
>
>
>
>
>
>
>
>
>
3=
>
7>
>
7>
>
7>
>
1 5>
>
>
;
mlf
11
***
2005 style
[21.12.20051:58pm]
[Invalid folder]
[162169]
14
v_yNS
15
vyNS ) vyNSd :
16
17
F:/T&F UK/Tcon/TCON_I_79_02/TCON_A_148880.3d
(TCON)
TCON_A_148880
168
6. Vehicle results
The following experiment was conducted on dry asphalt
with an instrumented prototype vehicle.
The manoeuvre is a lane change, conducted at
constant 100 kph longitudinal speed. This manoeuvre
leads the passive vehicle to spinning, while the
controlled vehicle is stable and tracks the target.
The lateral velocity used in the controller was
estimated via state observer.
Figure 4 shows the yaw rate tracking, which is not
ideal mainly due to model inaccuracies and to limitations from physics (one can see from gure 7 that the
lateral acceleration reaches its maximum). The yaw
rate is stable and regulated to the target.
The steering control action is shown in gure 5, it
results mainly in reducing the driver input to the front
tyre saturation value, while the brakes control action
7. Conclusion
A compact integrated steer and brakes vehicle controller, resulting from the application of state feedback
linearization technique and from additional vehicle
dynamics based choices was proposed, whose state
properties can be proved. The nal solution results
quite simple and reasonable from a vehicle dynamics
point of view.
The rst experimentation data showed that the controller was able to stabilize the vehicle in main working
conditions, allowing higher slip angle then with brakes
stand-alone systems.
FL
FR
RL
RR
target
meas
20
20
10
0
10
20
10
30
17
15
18
19
20
Figure 4.
21
22 23
time (s)
24
25
26
27
0
17
18
19
20
Figure 6.
21
22
time (s)
23
24
25
ay
beta
8
6
Lat Acc (m/s2), beta (deg)
5
Road Angle (deg)
27
Brake pressures.
10
meas
driver/VGR
ctrl
26
4
2
0
2
4
6
8
10
17
10
18
Figure 5.
***
2005 style
19
20
21
22
time (s)
23
24
25
26
27
[Invalid folder]
18
19
20
21
22
23
24
25
26
27
time (s)
[21.12.20051:58pm]
17
[162169]
Figure 7.
F:/T&F UK/Tcon/TCON_I_79_02/TCON_A_148880.3d
(TCON)
TCON_A_148880
169
References
J. Ackermann, Robust car steering by yaw rate control, CDC,
1990.
J. Ackermann, Robust control prevents car skidding, Bode Prize,
1996.
R. Busch, Seibertiz and P. Schmitz, IVDC the development of integrated vehicle dynamics control, VDI, 2003.
***
2005 style
[21.12.20051:58pm]
[Invalid folder]
[162169]
F:/T&F UK/Tcon/TCON_I_79_02/TCON_A_148880.3d
(TCON)
TCON_A_148880