applied

sciences
Article
Central Non-Linear Model-Based Predictive Vehicle
Dynamics Control
Philipp Maximilian Sieberg * and Dieter Schramm

Chair of Mechatronics, Faculty of Engineering, University of Duisburg-Essen, 47057 Duisburg, Germany;

[email protected]
* Correspondence: [email protected]; Tel.: +49-203-379-1862

Featured Application: This contribution presents a central predictive control of the vehicle dy-
namics regarding the roll, self-steering and pitch behavior.

Abstract: Considering automated driving, vehicle dynamics control systems are also a crucial
aspect. Vehicle dynamics control systems serve as an important influence factor on safety and
ride comfort. By reducing the driver’s responsibility through partially or fully automated driving
functions, the occupants’ perception of safety and ride comfort changes. Both aspects are focused
even more and have to be enhanced. In general, research on vehicle dynamics control systems
is a field that has already been well researched. With regard to the mentioned aspects, however,
a central control structure features sufficient potential by exploiting synergies. Furthermore, a
predictive mode of operation can contribute to achieve these objectives, since the vehicle can act
in a predictive manner instead of merely reacting. Consequently, this contribution presents a
central predictive control system by means of a non-linear model-based predictive control algorithm.





 In this context, roll, self-steering and pitch behavior are considered as control objectives. The
Citation: Sieberg, P.M.; Schramm, D.
active roll stabilization demonstrates an excellent control quality with a root mean squared error
Central Non-Linear Model-Based of 7.6953 × 10−3 rad averaged over both validation maneuvers. Compared to a vehicle utilizing a
Predictive Vehicle Dynamics Control. conventional control approach combined with a skyhook damping, pitching movements are reduced
Appl. Sci. 2021, 11, 4687. https:// by 19.75%. Furthermore, an understeering behavior is maintained, which corresponds to the self-
doi.org/10.3390/app11104687 steering behavior of the passive vehicle. In general, the central predictive control, thus, increases
both ride comfort and safety in a holistic way.
Academic Editors: Flavio Farroni,
Andrea Genovese and Keywords: central control; non-linear model-based predictive control; pitch behavior; predictive
Aleksandr Sakhnevych control; roll behavior; self-steering behavior; vehicle dynamics

Received: 24 April 2021

Accepted: 17 May 2021
Published: 20 May 2021
1. Introduction
Publisher’s Note: MDPI stays neutral
Two major driving factors in vehicle development are increasing the safety and en-
with regard to jurisdictional claims in
hancing the ride comfort of the vehicle [1]. Especially in the context of automated driving,
published maps and institutional affil- where the driver becomes a passenger, the perception of ride comfort changes significantly
iations. and at the same time gains in importance [2]. Moreover, a predictive mode of operation
of the vehicle is beneficial. The implementation of a central predictive control of the vehi-
cle dynamics addresses the objectives of increasing safety and ride comfort. The central
control structure exploits synergies in terms of the control quality [3]. An overview of the
Copyright: © 2021 by the authors.
state of the art for centralized integrated vehicle dynamics control systems is given in [4].
Licensee MDPI, Basel, Switzerland.
Furthermore, a predictive mode of operation allows the vehicle to act in a predictive way
This article is an open access article
instead of just reacting [5]. In addition to classical approaches to control vehicle dynamics
distributed under the terms and such as skyhook damping [6] and model-based control algorithms [7], the utilization of
conditions of the Creative Commons artificial intelligence is gaining increased attention. This mainly includes reinforcement
Attribution (CC BY) license (https:// learning [8], fuzzy inference systems [9] as well as deterministic artificial intelligence [10].
creativecommons.org/licenses/by/ Due to the current regulation for the use of artificial intelligence in vehicles, a model-based
4.0/). approach is considered in this contribution.

Appl. Sci. 2021, 11, 4687. https://doi.org/10.3390/app11104687 https://www.mdpi.com/journal/applsci

Appl. Sci. 2021, 11, 4687 2 of 18

For this purpose, a non-linear model-based predictive control algorithm implements

the central predictive control. The control objectives pursued are an active roll stabilization,
the manipulation of the self-steering behavior as well as the reduction of pitching move-
ments. Mathematical models are used to predict the system behavior as a function of the
manipulated variables [11]. Subsequently, the predicted system behavior is adapted to the
desired system behavior in the form of reference trajectories. Furthermore, the manipulated
variables can also be taken into account in the cost function to be minimized, so that the
energy requirement within the central predictive vehicle dynamics control can likewise be
reduced. A further advantage of the model-based predictive control is that it can also con-
sider constraints on the manipulated variables. This enables actuator limits to be respected
already during the optimization within the model-based predictive control algorithm [12].
As a result, this class of algorithms exhibits an excellent control quality. In [13], a model-
based predictive control is used to stabilize a vehicle at its vehicle dynamics limits. Due to
limitations of the side-slip angle and the yaw rate within the control algorithm the safety is
enhanced. [14] apply a model-based predictive control algorithm to reduce vertical vehicle
body motions. The vehicle is equipped with active suspension elements. Compared to a
passive vehicle the ride comfort is significantly increased. A model-based predictive control
algorithm is used in [15] to control the semi-active suspensions of a vehicle. This algorithm
is validated against classical control approaches such as the skyhook damping presented
in [16] and a clipped control strategy presented in [17]. For various road excitations, the
model-based predictive control algorithm outperforms the classical control approaches.
This contribution is organized as follows: Section 2 presents the simulation framework,
which is used to develop and validate the central predictive vehicle dynamics control.
Section 3 introduces the central non-linear model-based predictive control algorithm with
respect to the control objectives of roll, self-steering and pitch behavior. The central
predictive control is then validated in Section 4. The contribution concludes in Section 5
with a summary as well as an outlook on future research tasks.

2. Simulation Framework
A simulation framework is used to implement the central predictive vehicle dynamics
control and its validation. This framework is based on a co-simulation between IPG
CarMaker and MATLAB & Simulink. Figure 1 illustrates the simulation framework. The
multi-body simulation within the software IPG CarMaker is used for a realistic simulation
of the vehicle. In addition to this realistic representation of the vehicle and the vehicle
dynamics, IPG CarMaker also offers the possibility to edit and simulate the environment
as well as driver models. In the context of the contribution, a vehicle of the sport utility
vehicle class, a Lexus RX400h, is utilized. Due to the heightened center of gravity, this
class of vehicle features higher tendencies towards movements in terms of rolling and
pitching, which ultimately presents a more challenging task for the vehicle dynamics
control. In order to accomplish the control and to achieve the control objectives, the vehicle
is equipped with active stabilizers and semi-active dampers. The sensor equipment of the
vehicle in IPG CarMaker is based on a minimalistic configuration. Available measured
.
quantities are the longitudinal acceleration a x , the lateral acceleration ay , the yaw rate ψ, the
steering wheel angle δSW , the velocity v and the wheel speeds nij . Further fixed parameters
of the vehicle are listed in Table 1.
Appl.
Appl.Sci. 2021,11,
Sci.2021, 11,4687
x FOR PEER REVIEW 3 3ofof1819

IPG CarMaker

Reference Semi-Active
Trajectories Dampers

Control Active
Algorithm Stabilizers

State
Estimators
MATLAB & Simulink

Figure 1. Simulation Framework, © 2021 IEEE. Reprinted, with permission, from [18].
Figure 1. Simulation Framework, © 2021 IEEE. Reprinted, with permission, from [18].

Table 1. Vehicle Parameters.

The implementation of all algorithms is done in MATLAB & Simulink. In addition to
the central predictive control, this also includes the generation
Parameter Value of reference trajectories
Unit
representing the control targets,
Vehicle Body Mass
the simulation of the actuators
1820
with regard to
kg
a realistic
mapping, as well as the implementation
Track Width of state estimators,
1.538which estimate the
m states nec-
essary for the control not
Wheelbasedetermined by sensors. Examples
2.75 for the implementation
m of
these state estimatorsTires
are presented in [19–21]. 235/55R18 -
Distance of the Center of Gravity to the Front Axle 1.343 m
Distance
Table of the Center
1. Vehicle of Gravity to the Rear Axle
Parameters. 1.407 m
Height of the Center of Gravity 0.682 m
Parameter
Height of the Center of Pitching 0.3257 Value m Unit
Height of the Center of Rolling
Vehicle Body Mass 0.2826 1820 m kg
Moment of Inertia about the Lateral Axis 2654 kg m2
Track Width 1.538 m
Moment of Inertia about the Longitudinal Axis 760 kg m2
Wheelbase
Moment of Inertia about the Vertical Axis 2774 2.75 kg m2m
Tires 235/55R18 -
Distance of the Center of Gravity to the Front Axle 1.343 m
The implementation of all algorithms is done in MATLAB & Simulink. In addition
Distance of the Center of Gravity to the Rear Axle 1.407 m
to the central predictive control, this also includes the generation of reference trajectories
Height of the Center of Gravity 0.682 m
representing the control targets, the simulation of the actuators with regard to a realistic
mapping, as wellHeight
as of
thethe Center of Pitching
implementation 0.3257 estimate themstates
of state estimators, which
Height of the Center of Rolling 0.2826
necessary for the control not determined by sensors. Examples for the implementation m of
Moment of Inertia about the Lateral
these state estimators are presented in [19–21]. Axis 2654 kg m²
InMoment of Inertia
the following, theabout
focusthe Longitudinal
is on the central Axis 760
non-linear model-based kg m²con-
predictive
trol algorithm.
Moment of Inertia about the Vertical Axis 2774 kg m²
In thePredictive
3. Central following,Control
the focus is on the central non-linear model-based predictive control
algorithm.
The steps of prediction and subsequent optimization characterize the central predictive
control based on the non-linear model-based predictive control algorithm [22]. Within this
contribution, the integrated model-based predictive control presented in [5] is extended
and elaborated with respect to influencing the self-steering behavior. In this context, the
control of roll behavior features the highest priority. Influencing the self-steering behavior
and reducing pitching movements are subordinate control objectives.
Appl. Sci. 2021, 11, 4687 4 of 18

3.1. Prediction
Theoretical modeling is used to generate the prediction models of the vehicle dynamics
as a function of the manipulated variables. As a result, three interrelated prediction models
are determined, which are presented individually. The split is made in relation to the
control objectives of influencing the roll, the self-steering and the pitch behavior. In this
context, u1 and u2 represent the manipulated variables of the counter roll torques at the
front and rear axles, respectively. The variable u3 is the damping factor of the semi-active
damper at the front left, u4 the damping factor of the semi-active damper at the front right,
u5 the damping factor of the semi-active damper at the rear left and u6 the damping factor
of the semi-active damper at the rear right.

3.1.1. Roll Behavior

To build the prediction model for the roll behavior, the vehicle body is cut free in
the y − z plane. Subsequently, the principle of angular momentum is set up around the
vehicle’s roll center. The resulting equation can be transformed according to the roll
..
acceleration ϕ(k) at a certain time step k:
1
Jxx [hGR may cos ϕ ( k )+ hGR mg sin ϕ ( k ) − u1 ( k ) − u2 ( k )
−2 s2S,f cS,f + s2S,r cS,r sin ϕ(k)
 .
− (u3 (k) + u4 (k))s2D,f ϕ(k) cos ϕ(k) (1)
 .
− (u5 (k) + u6 (k))s2D,r ϕ(k) cos ϕ(k)]
..
= ϕ ( k ).

Here, Jxx represents the moment of inertia about the x-axis, hGR the distance between the
center of gravity and the roll center and m the mass of the vehicle body. The external input
variables are the lateral acceleration ay and the gravitational acceleration g. In addition to
the external input variables, the chassis elements also have an effect on the roll motion.
Apart from the active stabilizers and the semi-active dampers, the vehicle is equipped
with passive springs. These passive springs are characterized by the spring stiffnesses
cS,i . Furthermore, sS,i and sD,i indicate the distances of the springs and dampers from the
vehicle’s center plane. The index i indicates which vehicle axle is concerned.
.
Using the scheme of the semi-implicit Euler method [23], the roll rate ϕ(k + 1) and
the roll angle ϕ(k + 1) can be predicted as a function of the manipulated variables, starting
..
from the roll acceleration ϕ(k ):
. . ..
ϕ ( k + 1) = ϕ ( k ) + ϕ ( k ) tS , (2)
.
ϕ ( k + 1) = ϕ ( k ) + ϕ ( k + 1) tS . (3)
Here, tS denotes the fixed step size.

3.1.2. Self-Steering Behavior

The basis for the prediction of the self-steering behavior is the single-track model [24].
Here, the wheels of an axle are virtually combined for modeling. The single-track model
can be used to describe and predict the self-steering behavior and, thus, the response of the
vehicle to steering movements [25]. Within the control system, the self-steering gradient
SSG is used as the characteristic variable:
(αf (k) − αr (k))
SSG (k) = (4)
ay

The variables αf and αr are the slip angles at the front and rear axles, respectively. The
.
slip angles are dependent of the yaw rate ψ, the velocity v and the side-slip angle β. In
addition, the steering angle δ affects the slip angle at the front axle αf . The parameters
 Appl. Sci. 2021, 11, 4687 5 of 18

lf and lr represent the distance from the center of gravity to the front axle and the rear
axle, respectively.
. !
lf ψ(k ) + v sin β(k)
αf (k) = δ − arctan (5)
v cos β(k)
. !
−lr ψ(k) + v sin β(k)
αr (k) = −arctan (6)
v cos β(k)

Whereas the steering angle and the velocity are kept constant within the prediction, the
yaw rate and the side-slip angle are predicted. For this purpose, both Newton’s principle in
the lateral direction and the principle of angular momentum in the x − y plane are applied:




cos δ Fy,fl (k ) + Fy,fr (k) + Fy,rl (k) + Fy,rr (k) . .
− ψ ( k ) = β ( k ), (7)
mv cos β(k )

1


..
lf cos δ Fy,fl (k) + Fy,fr (k) − lr Fy,rl (k) + Fy,rr (k ) = ψ(k ). (8)
Jzz
. ..
These two equations are solved for β and ψ, respectively. Fy,fl , Fy,fr , Fy,rl and Fy,rr rep-
resent the lateral forces at the tire front left, front right, rear left and rear right, respectively.
The moment of inertia about the vertical axis is denoted as Jzz . By applying the scheme of
.
the explicit Euler integration method [26], the time derivative of the side-slip angle β(k)
..
and the yaw acceleration ψ(k) are then used to predict the side-slip angle β(k + 1) and the
.
yaw rate ψ(k + 1), respectively:
.
β ( k + 1) = β ( k ) + β ( k ) tS , (9)
. . ..
ψ ( k + 1) = ψ ( k ) + ψ ( k ) tS . (10)
The influence of the chassis elements and, thus, the actuators on the self-steering
behavior is exerted indirectly via the lateral tire forces Fy,ij . Here, the index j denotes the
vehicle side. The lateral tire forces Fy,ij correspond to the product of the slip angles αi and
the respective cornering stiffnesses cα,ij :

Fy,ij (k) = cα,ij (k)αi (k). (11)

Appl. Sci. 2021, 11, x FOR PEER REVIEW 6 of 19

The cornering stiffness cα,ij depends on the current wheel load Fz,ij . The dependency
features a degressive characteristic. This characteristic is illustrated in Figure 2.
,
Cornering Stiffness 𝑐

Wheel Load 𝐹 ,

Figure 2. Degressive
Figure Characteristic
2. Degressive of the
Characteristic of Cornering Stiffness
the Cornering Regarding
Stiffness the the
Regarding Wheel Load.
Wheel Load.

This relationship is modeled using a semi-empirical approach according to [27]. The

degressive characteristic is described by a mathematical model, which is parameterized
by empirical measurements:
𝐹, 𝑘
𝑐 , 𝑘 =𝑐 𝑐 𝐹 , sin 2 arctan . (12)
𝑐 𝐹 ,
The basis for the parameter identification is the tire model used in IPG CarMaker.
Appl. Sci. 2021, 11, 4687 6 of 18

This relationship is modeled using a semi-empirical approach according to [27]. The

degressive characteristic is described by a mathematical model, which is parameterized by
empirical measurements:
!!
Fz,ij (k)
cα,ij (k) = c1 c2 Fz0,ij sin 2arctan . (12)
c2 Fz0,ij

The basis for the parameter identification is the tire model used in IPG CarMaker.
This results in the parameters c1 and c2 of the semi-empirical tire model. Furthermore,
Fz0,ij is the nominal wheel load, which is present under static conditions. The influence
of the actuators and, thus, of the control is taken into account via the wheel loads Fz,ij .
The forces of the chassis elements are determined as a function of the roll behavior. The
parameter sSt,i indicates the distance of the stabilizer force application point from the
vehicle’s center plane:
 
lr . 1
Fz,fl (k ) = 0.5mg − u3 (k )sD,f ϕ(k) cos ϕ(k) − sS,f cS,f sin ϕ(k ) − u (k ) , (13)
lf + lr 2sSt,f 1
 
lr . 1
Fz,fr (k) = 0.5mg + u4 (k )sD,f ϕ(k) cos ϕ(k) + sS,f cS,f sin ϕ(k) + u (k) , (14)
lf + lr 2sSt,f 1
 
lf . 1
Fz,rl (k) = 0.5mg − u5 (k)sD,r ϕ(k ) cos ϕ(k) − sS,r cS,r sin ϕ(k) − u2 (k ) , (15)
lf + lr 2sSt,r
 
lf . 1
Fz,rr (k) = 0.5mg + u6 (k )sD,r ϕ(k) cos ϕ(k ) + sS,r cS,r sin ϕ(k) + u2 (k) . (16)
lf + lr 2sSt,r

3.1.3. Pitch Behavior

In order to obtain the prediction model for the pitch behavior, the vehicle body is
cut free in the x − z plane. Subsequently, the principle of angular momentum is set up
around the vehicle’s pitch center. The transformation of the resulting equation to the pitch
..
acceleration θ (k) yields:
 
1 2 2
Jyy lbrackh GP ma x cos θ ( k ) + h GP mg sin θ ( k ) − 2 lS, f cS, f + lS,r cS,r sin θ ( k )
l l
−u1 (k) sSt, f
St, f
− u2 (k) sSt,r
St,r
 . (17)
2
− (u3 (k) + u4 (k))l D, f θ (k) cos θ (k)
 . ..
2
− (u5 (k) + u6 (k))l D,r θ (k) cos θ (k)] = θ (k ).

Jyy denotes the moment of inertia about the lateral axis of the vehicle. The distance
between the center of gravity and the pitch center is defined by hGP . The parameters
lS,i , lD,i and lSt,i represent the distances between the center of gravity plane and the force
application points of the springs, dampers and stabilizers, respectively. Based on the pitch
.. .
acceleration θ (k) at time k, the pitch rate θ (k + 1) and the pitch angle θ (k + 1) for the time
k + 1 are likewise determined using the scheme of the semi-implicit Euler method:
. . ..
θ ( k + 1) = θ ( k ) + θ ( k ) tS , (18)
.
θ ( k + 1) = θ ( k ) + θ ( k + 1) tS . (19)
This procedure likewise allows the pitch behavior to be predicted as a function of the
manipulated variables.
Appl. Sci. 2021, 11, 4687 7 of 18

3.2. Optimization
Following the prediction of the vehicle dynamics as a function of the manipulated
variables, the optimization is executed with regard to the control objectives. The optimiza-
tion is performed using the entire prediction horizon np . The prediction horizon equals
0.15 s.
A major advantage of the non-linear model-based control algorithm is that constraints
can be taken into account within the optimization. In this contribution, the manipulated
variables are constrained. Thus, the physical limits of the actuators can be considered
within the optimization. This results in the restriction of the counter roll torques u1 and u2
between a minimum counter roll torque Tmin and a maximum counter roll torque Tmax

Tmin ≤ ui ≤ Tmax , i ∈ {1, 2}, (20)

as well as the restriction of the damping factors u3 , u4 , u5 and u6 in between a minimum

damping factor dmin and a maximum damping factor dmax

dmin ≤ ui ≤ dmax , i ∈ {3, 4, 5, 6}. (21)

For the description of the manipulated variables, temporal polynomials defined

over the prediction horizon are used to take into account the temporal course within the
optimization [5]. The definition of the polynomial degree is done with respect to the desired
characteristics. The manipulated variables of the counter roll torques at the front and rear
axle u1 and u2 , respectively, are defined as cubic polynomials:

u1 (k ) = a11 + a12 k + a13 k2 + a14 k3 , (22)

u2 (k ) = a21 + a22 k + a23 k2 + a24 k3 . (23)

The manipulated variables representing the damping factors u3 , u4 , u5 and u6 are
specified by quadratic polynomials:

u3 (k) = a31 + a32 k + a33 k2 , (24)

u4 (k) = a41 + a42 k + a43 k2 , (25)

u5 (k) = a51 + a52 k + a53 k2 , (26)
u6 (k) = a61 + a62 k + a63 k2 . (27)
For the further description, the following notation is used:

u(k ) = (u1 (k ), . . . , u6 (k))T , (28)

a = ( a11 , . . . , a63 )T . (29)

The manipulated variables are grouped in the vector u and the parameters of the
polynomials are grouped in the vector a. Furthermore, the predicted vehicle dynamic states
of the roll angle ϕ, the pitch angle θ and the self-steering gradient SSG are summarized in
the vector x:
x(k) = ( ϕ(k), θ (k), SSG (k ))T . (30)
The reference variables of the central predictive control are given in xRef . These result
from the generation of the reference trajectories:

xRef (k) = ( ϕRef (k ), θRef (k), SSGRef (k ))T . (31)

Appl. Sci. 2021, 11, 4687 8 of 18

A dynamic roll angle specification is used for the control objective of the active roll
stabilization. A non-linear roll model with passive chassis elements is used for this purpose:
1 2
Jxx[ h GR may cos ϕ p ( k )+ h GR mg sin ϕ p ( k ) − 2( sS, f cS, f + s2S,r cS,r ) sin ϕ p (k)
.
− d p, f s2D, f + d p,r s2D,r ϕ p (k) cos ϕ p (k)
c s
a  (32)
−2 St,b f St, f arcsin 2bSt, f sin ϕ p (k)
St, f  St, f ..

c sSt,r aSt,r
−2 St,r bSt,r arcsin 2bSt,r sin ϕ p ( k ) ] = ϕ p ( k ).

Here, dp,i represents the damping factors of the passive dampers. The passive stabi-
lizers are characterized by the stiffnesses cSt,i , the effective lengths aSt,i and the lever arms
..
bSt,i . The double integration of the passive roll acceleration ϕp by the explicit Euler method
yields the corresponding roll angle ϕp . This passive roll angle ϕp is then scaled by a scaling
factor ζ in order to determine the dynamic roll angle specification ϕRef :

ϕRef = ζ ϕp . (33)

This dynamic reference roll angle specification improves comfort and safety overall,
since the roll behavior is significantly reduced while still maintaining feedback of the lateral
dynamics to the driver. For the pitch and self-steering behavior, static reference values
are specified. The pitch angle specification θRef corresponds to the stationary pitch angle
of the vehicle. The specification of the self-steering gradient SSGRef is used to achieve an
understeering vehicle behavior corresponding to the passive vehicle behavior.
In addition to maintaining the reference trajectories, the optimization also takes into
account the energy requirements of the actuators, which should be set to a minimum. For
this reason, the weighting factors λ are introduced within the optimization. Thus, the focus
can be set on the control quality as well as on the energy demand:

λ = (λR , λP , λS , λu1 , λu2 , λu3 , λu4 , λu5 , λu6 )T . (34)

During the optimization, the cost function f (u(k), x(k ), xRef (k)) is minimized for
the entire prediction horizon np by adjusting the parameters of the polynomials a. The
optimization toolbox using the interior-point algorithm of MATLAB is used to solve the
optimization [28,29]:
 
min f u(k), x(k), xRe f (k) =
a
np  2
1
np [ λ R ∑ ϕ Re f ( k ) − ϕ ( k )
k =0
np  2 np  2
+λS ∑ SSGRe f (k) − SSG (k) + λ P ∑ θ Re f (k) − θ (k) (35)
k =0 k =0
np np np
+ λu1 ∑ (u1 (k)) + λu2 ∑ (u2 (k)) + λu3 ∑ (u3 (k))2
2 2
k =0 k =0 k =0
np np np
+ λu4 ∑ (u4 (k)) + λu5 ∑ (u5 (k)) + λu6 ∑ (u6 (k))2 ].
2 2
k =0 k =0 k =0

The result of the optimization is a set of optimal polynomial parameters over the entire
prediction horizon, from which optimal manipulated variable trajectories are obtained.
Finally, the principle of the receding horizon is applied [22]. Only the manipulated variables
for the next time step are taken from the optimal manipulated variable trajectories and
passed on to the actuator models. In the next time step, the entire process of prediction
and optimization is repeated. This allows the non-linear model-based predictive control
to adapt to non-modeled disturbances in an optimal way. Furthermore, the warm-start
method is used [30], in which the last determined optimal polynomial parameters are used
as a starting point for the optimization in the following time step. As a result, the number
of iterations within the optimization can be reduced.
Appl. Sci. 2021, 11, 4687 9 of 18

4. Results
In a first step, the validation maneuvers are presented which are used to validate the
central predictive vehicle dynamics control. Subsequently, the focus is on the evaluation of
the control quality for the individual driving maneuvers. The central predictive control is
evaluated against a vehicle using a conventional roll control as well as a skyhook damping
according to [16] and a vehicle with passive chassis elements. The conventional roll control
is based on a proportional integral derivative controller, which is parameterized by using
the control system design toolbox of MATLAB. The section concludes with a summary of
the results obtained.

4.1. Validation Maneuvers

In order to validate the central non-linear model-based predictive control algorithm,
two driving maneuvers are utilized. First, the double lane change driving maneuver is
used [31]. For this purpose, the vehicle is first accelerated from standstill to the target
velocity before performing the double lane change at constant velocity. The first lane change
is performed counterclockwise. The target velocity within this contribution equals 50 km/h.
The track limits for the double lane change are defined in the ISO standard. By combining
the acceleration phase and the dynamic lane changes, the central predictive control can be
validated with regard to all control objectives. The second driving maneuver for validation
is a sinusoidal steering. This maneuver also involves accelerating the vehicle to a velocity
of 50 km/h and then performing sinusoidal steering, according to [32]. Three periods
with a steering wheel angle amplitude of 68◦ and a frequency of 1 Hz are performed. In
addition, a significantly reduced friction coefficient of 0.4 is used for this driving maneuver.

4.2. Double Lane Change

The evaluation of the control quality for the double lane change is carried out sepa-
rately for each control objective. The evaluation is done qualitatively and quantitatively.

4.2.1. Roll Behavior

For the qualitative evaluation of the control quality of the non-linear model-based
predictive control algorithm with respect to the main control objective, the roll angle is
plotted over time for the validation maneuver. This is illustrated in Figure 3. The roll
angle ϕMpc resulting from the execution of the non-linear model-based predictive control
algorithm is represented by a dotted black line. The dynamic reference variable ϕRef is
represented by a red solid line. Furthermore, a green dashed line illustrates the roll angle
curve ϕPas of the vehicle with passive chassis elements and an orange dotted line illustrates
the roll angle curve ϕPID resulting from the conventional control approach.
The reference variable ϕRef reduces the roll motion by about 75% compared to the
passive vehicle ϕPas . The roll angle resulting by the central predictive control follows
the reference with an excellent accuracy. In contrast, the conventional control approach
based on the proportional integral derivate controller and the skyhook damping results in
increased control deviations. Thus, with the central predictive control, not only the safety
but also the comfort is increased compared to the conventional control approach as well as
to the passive vehicle behavior. The rolling movements that the passive vehicle exhibits
during acceleration are not present for the vehicle utilizing the central predictive control
and the vehicle utilizing the conventional control approach, respectively.
 plotted over time for the validation maneuver. This is illustrated in Figure 3. The roll a
𝜑 resulting from the execution of the non-linear model-based predictive control a
rithm is represented by a dotted black line. The dynamic reference variable 𝜑 is re
sented by a red solid line. Furthermore, a green dashed line illustrates the roll angle c
𝜑 of the vehicle with passive chassis elements and an orange dotted line illustrate
Appl. Sci. 2021, 11, 4687 10 of 18
roll angle curve 𝜑 resulting from the conventional control approach.

𝜑 𝜑 𝜑 𝜑

0.02

Roll Angle in rad

0.01

0
Appl. Sci. 2021, 11, x FOR PEER REVIEW 11 of 1
-0.01

Furthermore, the root mean squared errors are determined for the quantitative eval
-0.02
uation:
𝟐
0 10 20
∑
30 𝑥 40 𝑘 − 𝑥 50 𝑘 60 70 (36
𝑅𝑀𝑆𝐸 𝑥 ,𝑥 = , 𝑖 ∈ PID, Mpc .
𝑛
Time in s
In relation to the active roll stabilization, the non-linear model-based predictive con
Figure 3. Control aQuality
trol exhibits root Double
mean Lane Change–Roll
squared of 2.3906 ×
error Behavior–Roll 10 Curves.
Angle rad. This corresponds t
Figure
0.0137°. 3. Control
In contrast, Quality Double Lane
the conventional Change–Roll
control Behavior–Roll
approach Angle
results Curves.
in aare
root mean squared
In addition, the impacts of the active roll stabilizations on the comfort evaluated
error of 0.0010
by examining the The rad, which
resulting is
frequencyequivalent
spectra. to 0.0573°. The quantitative evaluation, thus
reference variable 𝜑 These
reducesarethe
illustrated in Figure
roll motion 4. Due
by about to compared
75% the to
confirms
use the
of the centralqualitative
vehicleanalysis.
predictive
passive control,
𝜑 . The aThe control
stronger
roll angle quality
damping
resulting byof
within
thethe
the central
frequency
central predictive
spectrum
predictive control
control followi
excellent.
is Inreference
general,
present compared thepassive
towith
the central
an predictive
vehicle.
excellent control,
The frequency
accuracy. thus,
In contrast, the outperforms
spectrum the conventiona
of the conventional
conventional control approach b
approach also
control approach.features a stronger damping than the passive vehicle. Compared
on the proportional integral derivate controller and the skyhook damping to the results i
central predictive control
creased algorithm,
control however,
deviations. Thus,awith
weaker
the damping is present.control, not only the s
central predictive
but also the comfort is increased compared to the conventional control approach as
as to the passive vehicle behavior. The rolling movements that the passive vehicle exh
𝜑
during acceleration are not present for𝜑the vehicle𝜑utilizing the central predictive co
and the vehicle utilizing the conventional control approach, respectively.
In addition, the impacts of the active roll stabilizations on the comfort are evalu
Amplitude in rad

0.002 by examining the resulting frequency spectra. These are illustrated in Figure 4. Due t
use of the central predictive control, a stronger damping within the frequency spect
is present compared to the passive vehicle. The frequency spectrum of the conventi
approach also features a stronger damping than the passive vehicle. Compared to the
tral predictive control algorithm, however, a weaker damping is present.
0.001

0
0 1 2 3 4 5 6 7 8 9 10
Frequency in Hz

Figure 4. Control Quality Double Lance Change-Roll Behavior-Frequency Spectra.

Figure 4. Control Quality Double Lance Change-Roll Behavior-Frequency Spectra.
Furthermore, the root mean squared errors are determined for the quantitative evaluation:
4.2.2. Self-Steering Behaviors
n

2
In comparison ∑k=1the
to controlling xRef
roll − xMpcinfluencing
(k)angle, (k) the self-steering behavior is
RMSE( xRef , xi ) = , i ∈ {PID, Mpc}. (36)
n
subordinate control objective. Here, a constant understeering behavior 𝑆𝑆𝐺 is pursued
which corresponds approximately to the one of passive vehicle. Because the vehicle i
equipped with active stabilizers and semi-active dampers, this control objective can only
be pursued to a limited extent. With regard to the representability for the evaluation, th
pseudo quantity 𝛼 is introduced:
 Appl. Sci. 2021, 11, 4687 11 of 18

In relation to the active roll stabilization, the non-linear model-based predictive control
exhibits a root mean squared error of 2.3906 × 10−4 rad. This corresponds to 0.0137◦ .
In contrast, the conventional control approach results in a root mean squared error of
0.0010 rad, which is equivalent to 0.0573◦ . The quantitative evaluation, thus, confirms the
qualitative analysis. The control quality of the central predictive control is excellent. In
general, the central predictive control, thus, outperforms the conventional control approach.

4.2.2. Self-Steering Behavior

In comparison to controlling the roll angle, influencing the self-steering behavior is a
subordinate control objective. Here, a constant understeering behavior SSGRef is pursued,
which corresponds approximately to the one of passive vehicle. Because the vehicle is
equipped with active stabilizers and semi-active dampers, this control objective can only
be pursued to a limited extent. With regard to the representability for the evaluation, the
pseudo quantity eαi is introduced:

αi = αf,i − αr,i = ay SSGi , ie{Ref, Pas, PID, Mpc}

e (37)

The pseudo quantity e αi represents the difference between the slip angle of the front
axle αf,i and rear axle αr,i and, thus, corresponds to the product of the self-steering gradient
Appl. Sci. 2021, 11, x FOR PEER REVIEW 12 of 19
SSGi and the lateral acceleration ay . The evaluation is conducted in the following based on
the pseudo quantity e αi . The qualitative evaluation is performed using the pseudo quantity
courses for the validation maneuver. This is shown in Figure 5.

0.004
𝛼 𝛼 𝛼 𝛼

0.002
Pseudo Quantity in rad

-0.002

-0.004

40 45 50 55 60 65 70
Time in s
Figure 5. Control Quality Double Lane Change–Self-Steering Behavior–Pseudo Quantity Curves.
Figure 5. Control Quality Double Lane Change–Self-Steering Behavior–Pseudo Quantity Curves.
With regard to the absence of lateral dynamics in the acceleration phase, only the
With regard to the absence of lateral dynamics in the acceleration phase, only the
section of the double lane change is considered. The representation of the pseudo quantity
section of the double lane change is considered. The representation of the pseudo quantity
suggests
suggests a dynamic
a dynamicreference variable𝛼e
referencevariable αRef.. However,
However, this this results
results from
fromthethedynamics
dynamicsofofthe
thelateral
lateralacceleration
accelerationay .𝑎With
. With regard
regard to safety,
to safety, a constant
a constant self-steering
self-steering SSGRef𝑆𝑆𝐺
gradient
gradient is used.
is The
used.representation
The representation
remains consistent with the evaluation of the roll behavior. behavior.
remains consistent with the evaluation of the roll The vehicle
The vehicle
with with the
the central centralcontrol
predictive predictive
shows control shows
an almost an almost
identical identical
course course quantity
of the pseudo of the
pseudo quantity to the passive vehicle. Due to the weighting of this control target
to the passive vehicle. Due to the weighting of this control target and the limited possibility and the
limited possibilitythere
to manipulate, to manipulate, therefrom
is a deviation is a the
deviation
referencefrom the reference
variable. variable.
Since the Since
conventional
thecontrol
conventional
approach control
does approach doesconsider
not explicitly not explicitly consideronthe
the influence theinfluence on the
self-steering self-
behavior,
steering behavior, only a limited evaluation can be performed. For the
only a limited evaluation can be performed. For the validation maneuver of the double validation maneu-
verlane
of the double
change, thelane change, the
conventional conventional
control exhibits a control exhibits
self-steering a self-steering
behavior behavior
corresponding to the
corresponding to the passive self-steering behavior, similar
passive self-steering behavior, similar to the central predictive control.to the central predictive con-
trol.
In principle, the control objective is satisfied by the central predictive control, since
the self-steering behavior corresponds to the behavior of the passive vehicle despite uti-
lizing an active control system. This is confirmed by examining the frequency spectra,
which are illustrated in Figure 6.
The frequency spectrum for the vehicle with the central predictive control is almost
 Appl. Sci. 2021, 11, 4687 12 of 18

Appl. Sci. 2021, 11, x FOR PEER REVIEW In principle, the control objective is satisfied by the central predictive control, since the
13 of 19
self-steering behavior corresponds to the behavior of the passive vehicle despite utilizing
an active control system. This is confirmed by examining the frequency spectra, which are
illustrated in Figure 6.

𝛼 𝛼 𝛼
0.0006
Amplitude in rad

0.0003

0
0 1 2 3 4 5 6 7 8 9 10
Frequency in Hz
Figure 6. 6.
Figure Control Quality
Control Double
Quality Lane
Double Change–Self-Steering
Lane Behavior–Frequency
Change–Self-Steering Spectra.
Behavior–Frequency Spectra.

The frequency
4.2.3. Pitch Behavior spectrum for the vehicle with the central predictive control is almost
identical to
Similar tothat
theofmanipulation
the passive vehicle.
of theThe conventional
self-steering controlthe
behavior, approach features
reduction a similar
of pitching
frequency is
movements spectrum.
a subordinate control objective within the non-linear model-based predic-
The quantitative
tive control algorithm. Due evaluation
to the for the whole
vehicle’s validation
equipment maneuver
with active results in and
stabilizers a root mean
semi-
squared error for the central predictive control of 8.1009 × 10 −4 rad, which is equiva-
active dampers,◦only a limited influence on the pitch behavior is possible. In order to eval-
lentthe
uate tocentral
0.0464 predictive
. The conventional
control in acontrol approach
qualitative yields
way with a roottomean
respect squared
the control error of
objective
8.1334 × 10 −4 rad, which corresponds to 0.0466◦ .
of reducing the pitching movements, the pitch angle is plotted over time. The pitch angle
curves
4.2.3. are shown
Pitch in Figure 7.
Behavior
Similar to the manipulation of the self-steering behavior, the reduction of pitching
movements is a subordinate control objective within the non-linear model-based predictive
control algorithm. Due to the vehicle’s equipment with active stabilizers and semi-active
0.04 𝜃 𝜃 𝜃 𝜃
dampers, only a limited influence on the pitch behavior is possible. In order to evaluate
the central predictive control in a qualitative way with respect to the control objective of
Pitch Angle in rad

reducing the pitching movements, the pitch angle is plotted over time. The pitch angle
0.02
curves are shown in Figure 7.
The focus is on the acceleration phase of the validation maneuver, since pitching move-
0 ments are present here due to the gear changes and the present longitudinal accelerations.
The pitching movements during the double lane change in contrast are negligibly small.
The representation remains consistent. Analogous to the manipulation of the self-steering
-0.02 behavior, a constant reference variable θRef is used. This reference variable represents the
stationary pitch angle of the vehicle, which corresponds to the pitch angle that is present
-0.04 when the vehicle is at standstill. The reference variable cannot be adjusted fully by the
present vehicle setup, but it is considered in order to implement the reduction of pitching
0 movements. In comparison 10 to the vehicle with passive chassis20elements θPas , pitching
movements of the vehicle with the central predictive control θMpc are, therefore, reduced.
The implementation of theTime in s damping within the conventional control approach
skyhook
also reduces pitching movements compared to the passive vehicle. By exploiting synergies,
however, the implementation of the central predictive control results in a greater reduction
Figure 7. Control Quality
of pitching Double Lane
movements Change–Pitch
compared Behavior–Pitch
to the conventionalAngle Curves.
control approach.

The focus is on the acceleration phase of the validation maneuver, since pitching
movements are present here due to the gear changes and the present longitudinal accel-
erations. The pitching movements during the double lane change in contrast are negligi-
bly small. The representation remains consistent. Analogous to the manipulation of the
 movements is a subordinate control objective within the non-linear model-base
tive control algorithm. Due to the vehicle’s equipment with active stabilizers a
active dampers, only a limited influence on the pitch behavior is possible. In orde
uate the central predictive control in a qualitative way with respect to the control
Appl. Sci. 2021, 11, 4687
of reducing the pitching movements, the pitch angle is plotted over time. The pi
13 of 18
curves are shown in Figure 7.

Appl. Sci. 2021, 11, x FOR PEER REVIEW 14 of 19

0.04 𝜃 𝜃 𝜃 𝜃
self-steering behavior, a constant reference variable 𝜃 is used. This reference variable
represents the stationary pitch angle of the vehicle, which corresponds to the pitch angle

Pitch Angle in rad

0.02
that is present when the vehicle is at standstill. The reference variable cannot be adjusted
fully by the present vehicle setup, but it is considered in order to implement the reduction
of pitching movements. In comparison to the vehicle with passive chassis elements 𝜃 ,
0
pitching movements of the vehicle with the central predictive control 𝜃 are, therefore,
reduced. The implementation of the skyhook damping within the conventional control
approach also-0.02
reduces pitching movements compared to the passive vehicle. By exploit-
ing synergies, however, the implementation of the central predictive control results in a
greater reduction of pitching movements compared to the conventional control approach.
-0.04
The corresponding frequency spectra are shown in Figure 8. Here, the positive influ-
ence of the central predictive control is also evident. The non-linear model-based predic-
0 10 20
tive control results in greater damping within the frequency spectrum. The amplitudes of
the conventional control approach present at very Timelowinfrequencies
s are sometimes even
greater than those of the passive vehicle. In terms of ride comfort, the central predictive
Figure 7. Control Quality Double Lane Change–Pitch Behavior–Pitch Angle Curves.
control, thus, provides a significant improvement in comparison to the application of the
conventional control
Figure
The correspondingapproach.
7. Control Quality Double
frequency spectraLane
areChange–Pitch
shown in Figure Behavior–Pitch
8. Here, theAngle Curves.
positive influ-
The quantitative evaluation confirms this statement. The non-linear
ence of the central predictive control is also evident. The non-linear model-based predictivemodel-based
predictive control in
control results yields aThe
greater rootfocus
meanissquared
damping on
within error of 0.0106
the acceleration
the frequency phase
rad,ofwhich
spectrum.the validation
corresponds
The maneuver,
amplitudes to of since
0.6073°. movements
The conventional
the conventional control controlare present
approach
approach present here
results due to
in a low
at very the gear changes
root frequencies
mean squared and
areerror of 0.0119
the
sometimespresent
even longitudi
0.0119 radthan
greater thoseerations.
corresponding
of theto 0.6818°.
The
passivepitching
By movements
vehicle. using ofduring
the central
In terms the double
predictive
ride comfort, lane change
thecontrol,
central in contrast ar
pitching
predictive
movements
control, thus, thus,bly
are, provides small.
reduced The representation
compared
a significant remains
to the conventional
improvement consistent. Analogous
controltoapproach
in comparison the to the
as well
application manipulati
ofasthe
compared to the passive vehicle.
conventional control approach. As a result, the ride comfort is, thus, increased.

𝜃 𝜃 𝜃

0.006
Amplitude in rad

0.003

0
0 1 2 3 4 5 6 7 8 9 10
Frequency in Hz

Figure 8. Control
Figure Quality
8. Control Double
Quality Lane
Double Change–Pitch
Lane Behavior-Frequency
Change–Pitch Spectra.
Behavior-Frequency Spectra.

The quantitative
4.3. Sinusoidal Steering evaluation confirms this statement. The non-linear model-based
predictive control
The evaluation ofyields a rootquality
the control mean for
squared error of steering
the sinusoidal 0.0106 rad, which is
maneuver corresponds
likewise
◦
to 0.6073qualitatively
. The conventional control approach results in phase,
a root mean squared error
performed and quantitatively. The acceleration which is already ex- of
0.0119 rad corresponding to 0.6818◦ . By using the central predictive control, pitching
amined in Section 4.2, is neglected and only the sinusoidal steering phase is considered.
movements are, thus, reduced compared to the conventional control approach as well as
compared to the passive vehicle. As a result, the ride comfort is, thus, increased.
 Appl. Sci. 2021, 11, 4687 14 of 18

4.3. Sinusoidal Steering

Appl. Sci. 2021, 11, x FOR PEER REVIEW The evaluation of the control quality for the sinusoidal steering maneuver is 15 of 19
likewise
performed qualitatively and quantitatively. The acceleration phase, which is already
examined in Section 4.2, is neglected and only the sinusoidal steering phase is considered.
4.3.1. Roll Behavior
4.3.1. Roll Behavior
To evaluate the control quality of the central predictive control with a focus on the
To evaluate the control quality of the central predictive control with a focus on the
active roll stabilization, the roll angle curves during the sinusoidal steering are plotted.
active roll stabilization, the roll angle curves during the sinusoidal steering are plotted. The
The present roll angle curves are shown in Figure 9.
present roll angle curves are shown in Figure 9.

𝜑 𝜑 𝜑 𝜑
0.06
Roll Angle in rad

-0.06

10 15 20
Time in s

Figure 9. Control Quality Sinusoidal Steering–Roll Behavior–Roll Angle Curves.

Figure 9. Control Quality Sinusoidal Steering–Roll Behavior–Roll Angle Curves.
As with the double lane change maneuver, the reference variable ϕRef reduces the
As with
rolling the double
motions lane change 75%
by approximately maneuver,
compared the to
reference motions 𝜑
rolling variable ϕPas ofreduces the
the passive
rolling motions by approximately 75% compared to rolling motions 𝜑
vehicle. The central predictive control ϕMpc is able to follow the reference variable ϕRef of the passive
vehicle.
despiteThe
thecentral
minimal predictive control 𝜑 Compared
friction coefficient. is able toto follow the reference
the results 𝜑
variablemaneu-
for the validation
despite
ver ofthe
theminimal friction
double lane coefficient.
change, Compared
there are slightly to the results
larger controlfor the validation
deviations due to maneu-
this low
verfriction
of the double laneIn
coefficient. change, there are
comparison slightly
to the larger control
conventional deviations
control approach, due
thetocentralized
this low
friction coefficient. In comparison to the conventional
predictive control nevertheless features an improved control approach,
quality. the centralized
predictive
Thiscontrol nevertheless
is confirmed by thefeatures
evaluationan improved
of the rootcontrol quality. errors. The non-linear
mean squared
This is confirmed
model-based by the
predictive evaluation
control yieldsofa the
rootroot
meanmean squared
squared errorerrors. The non-linear
of 0.0013 rad for the
model-based predictive
sinusoidal steering, control
which yields a root
is equivalent mean ◦squared
to 0.0745 errorthe
. In contrast, 0.0013 rad for
of conventional the
control
sinusoidal
approachsteering,
results inwhich
a rootismean
equivalent
squared 0.0745°.
toerror In contrast,
of 0.0031 the conventional
rad, which corresponds to 0.1776◦ .
control
approach results
Despite the in a root
aggravated mean
road squared
conditions, theerror
control 0.0031 ofrad,
of quality the which
central corresponds to
predictive control
0.1776°. Despite the aggravated road conditions, the control quality of the central predic-
is excellent.
tive control is excellent.
4.3.2. Self-Steering Behavior
4.3.2. Self-Steering Behavior
For the evaluation of the control quality with respect to the self-steering behavior,
theFor
courses
the evaluation of thequantity
of the pseudo αi are first
control quality
e withconsidered again.
respect to the The resulting
self-steering curves
behavior, theare
illustrated in Figure 10.
courses of the pseudo quantity 𝛼 are first considered again. The resulting curves are il-
lustratedComparable
in Figure 10.to the results for the double lane change, the self-steering behavior
αMpc
e implemented
Comparable to thebyresults
the central
for thepredictive
double lane control
change,system almost corresponds
the self-steering behavior 𝛼to the
passive self-steering behavior α Mpc . With regard to the reference
implemented by the central predictive control system almost corresponds toe
e variable Ref ,passive
αthe there are
larger deviations due to the very low friction coefficient. Through the
self-steering behavior 𝛼 . With regard to the reference variable 𝛼 , there are largerlimited adjustment
possibilities caused by the vehicle equipment and the prioritization of the control objectives,
deviations due to the very low friction coefficient. Through the limited adjustment possi-
only a limited improvement of the control quality is achieved by the central predictive
bilities caused by the vehicle equipment and the prioritization of the control objectives,
control compared to the passive vehicle. A similar self-steering behavior is obtained by
only a limited improvement of the control quality is achieved by the central predictive
control compared to the passive vehicle. A similar self-steering behavior is obtained by
the conventional control approach, though slightly larger deviations from the reference
variable are present.
 Appl. Sci. 2021, 11, 4687 15 of 18

Appl. Sci. 2021, 11, x FOR PEER REVIEW

Appl. Sci. 2021, 11, x FOR PEER REVIEW 16 of 19

the conventional control approach, though slightly larger deviations from the reference
variable are present.

𝛼 𝛼 𝛼 𝛼 𝛼 𝛼 𝛼 𝛼
0.03 0.03
Pseudo Quantity in rad

Pseudo Quantity in rad

0 0

-0.03 -0.03

10 10 15 15 20 20
Time in s Time in s
Figure 10. Control Quality Sinusoidal
Figure Steering–Self-Steering
10. Control Quality Sinusoidal Behavior–Pseudo Quantity
Steering–Self-Steering Curves.
Behavior–Pseudo Quantity Curves.
Figure 10. Control Quality Sinusoidal Steering–Self-Steering Behavior–Pseudo Quantity Curves.
The non-linear model-based predictive
The non-linear control algorithm
model-based predictiveresults
controlinalgorithm
a root mean results in a r
The non-linear model-based predictive control algorithm results in a root mean
squared error of 0.0077 rad error
squared of 0.0077 to
corresponding rad0.4412 °. The
corresponding to 0.4412
conventional control
°. The ap-
conventional c
squared error of 0.0077 rad corresponding to 0.4412◦ . The conventional control approach
proach yields a root mean
proach squared
yields error
a root of 0.0078
mean squared whichofis0.0078
rad, error equivalent
rad, to 0.4469°.
which is equivalent to
yields a root mean squared error of 0.0078 rad, which is equivalent to 0.4469◦ .
In general, the controlIn objective
general, of influencing the self-steering behavior is achievedbehavior i
In general, the control objectivethe
of control objective
influencing of influencing
the self-steering the self-steering
behavior is achieved as
as well. A self-steering behavior corresponding to thecorresponding
passive self-steering behavior is
well. A self-steeringasbehavior
well. A corresponding
self-steering behavior
to the passive to behavior
self-steering the passive self-steering
is present b
present despite the application
present of the
despite central
the predictive
application of control
the
despite the application of the central predictive control system. centralsystem.
predictive control system.

4.3.3. Pitch
4.3.3. Behavior
Pitch Behavior 4.3.3. Pitch Behavior
The qualitative
The qualitativeevaluation
evaluation for
forthe
thecontrol
The qualitative objective
evaluation
control of
ofreducing
for the
objective pitching
control objective
reducing movements
pitching of reducing pitching
movements is m
is likewise
likewise realized
realizedbyby examining
is examining
likewise the
realized pitch
by angle curves
examining theduring
pitch the sinusoidal
angle curves steering.
during
the pitch angle curves during the sinusoidal steering. Thethe sinusoida
The pitch
pitch angle
angle curves
curves areare
The shown
pitch
shown in in
angle Figure
curves
Figure 11.11.
are shown in Figure 11.

0.01 0.01
Pitch Angle in rad

Pitch Angle in rad

𝜃 𝜃 𝜃𝜃 𝜃𝜃 𝜃 𝜃
0 0
10 10 15 15 20
Time in s Time in s
Figure
Figure 11.11. Control
Control Quality
Quality Sinusoidal
Sinusoidal
Figure Steering–Pitch
Steering–Pitch
11. Control Behavior–Pitch
Behavior–Pitch
Quality Sinusoidal Angle
Angle
Steering–Pitch Curves.
Curves.
Behavior–Pitch Angle Curves.

The
The positive
positive influence
influence ofof
The centralpredictive
central
positive predictiveof
influence control
control isispredictive
central likewise evident
likewise evident
controlduring
during the
the sinu-
is likewise si-
evident dur
soidal steering. Compared
nusoidal steering. Compared to the passive
to the passive
nusoidal steering. vehicle and
vehicle
Compared the conventional
and passive
to the control
the conventional
vehicle and approach,
control ap-
the conventional c
the pitching
proach, movements
the pitching are
movements reduced.
are reduced.
proach, the pitching movements are reduced.
Appl. Sci. 2021, 11, 4687 16 of 18

The quantitative evaluation confirms these results. Whereas the vehicle utilizing the
conventional control approach deviates from the stationary pitch angle θRef by a root mean
squared error of 0.0035 rad, which corresponds to 0.2005◦ , the non-linear model-based
predictive control results in a root mean squared error of 0.0025 rad, which corresponds
to 0.1432◦ .
In comparison to the passive vehicle and the application of the conventional con-
trol approach, the utilization of the central predictive control system, thus, increases the
ride comfort.

4.4. Conclusions
The central predictive vehicle dynamics control fulfills the defined control objectives
as desired. A summary of the control quality is given in Table 2. In addition to this, the
results are compared with those of a vehicle with passive chassis elements and those
obtained by a conventional control approach.

Table 2. Central Predictive Vehicle Dynamics Control—Root Mean Squared Errors.

Setup Driving Maneuver Vehicle Dynamics RMSE Unit

Roll Behavior 0.0043 rad
Double Lane Change Self-Steering Behavior 8.2703 × 10−4 rad
Pitch Behavior 0.0114 rad
Passive Chassis
Roll Behavior 0.0099 rad
Sinusoidal Steering Self-Steering Behavior 0.0080 rad
Pitch Behavior 0.0034 rad
Roll Behavior 0.0010 rad
Double Lane Change Self-Steering Behavior 8.1334 × 10−4 rad
Proportional Integral
Pitch Behavior 0.0119 rad
Derivative Control and
Roll Behavior 0.0031 rad
Skyhook Damping
Sinusoidal Steering Self-Steering Behavior 0.0078 rad
Pitch Behavior 0.0035 rad
Roll Behavior 2.3906 × 10−4 rad
Double Lane Change Self-Steering Behavior 8.1009 × 10−4 rad
Central Predictive Pitch Behavior 0.0106 rad
Control Roll Behavior 0.0013 rad
Sinusoidal Steering Self-Steering Behavior 0.0077 rad
Pitch Behavior 0.0025 rad

The main control objective of the central predictive vehicle dynamics control, the active
roll stabilization, is met with an excellent control quality. In comparison to the conventional
control approach, the control quality is improved by approximately 67.08%. Likewise, good
results are achieved with regard to the further subordinate control objectives. As a result,
pitching movements are reduced by approximately 19.75% compared to the conventional
control approach. Despite the intervention of the control system, the self-steering behavior
of the controlled vehicle corresponds to that of the passive vehicle.
The validation demonstrates the advantages of a centralized predictive control struc-
ture over a passive vehicle and a vehicle using a conventional control approach comprising
a proportional integral derivative controller and a skyhook damping. The improvements in
the control quality by of a non-linear model-based predictive control compared to a linear
control approach utilizing a skyhook damping for a pure active roll stabilization are also
presented in [33].

5. Conclusions and Outlook

This contribution presents a central predictive vehicle dynamics control system. The
control objectives of an active roll stabilization, a manipulation of the self-steering be-
havior as well as a reduction of pitching movements are pursued. The control system is
Appl. Sci. 2021, 11, 4687 17 of 18

implemented by a non-linear model-based predictive control algorithm. The simulation

framework comprising a co-simulation between IPG CarMaker and MATLAB & Simulink
is used. The test vehicle utilized is a sport utility vehicle equipped with active stabilizers
and semi-active dampers to realize the control. Due to the heightened center of gravity,
the sport utility vehicle features an increased challenge in influencing rolling and pitch-
ing motions. Here, the implementation of active roll stabilization is the primary control
objective. The weighting of the control objectives is done via corresponding weighting
factors within the optimization of the non-linear model-based predictive control algorithm.
The validation is performed for the driving maneuvers of the double lane change and the
sinusoidal steering. In conclusion, the implementation of the central predictive control
demonstrates an excellent control quality. The central predictive control increases safety
and comfort significantly compared to a vehicle with passive chassis elements and a vehicle
utilizing a conventional control approach.
The focus of this contribution is on the presentation of the central predictive vehicle
dynamics control based on the non-linear model-based predictive control algorithm and
the resulting excellent control quality. However, due to the numerical solution of the
optimization problem, an increased computational effort is imposed at the same time. With
regard to a future implementation of the central predictive control, a real-time implementa-
tion of this central predictive vehicle dynamics control with a reduced computational effort
is required. Whereas classical approaches linearize the underlying prediction models and,
thus, simplify them, making them analytically solvable, or limit the iteration steps of the
optimization, future work will explore the use of artificial intelligence. Initial conceptual
research has already shown a particular suitability of neuro-fuzzy inference systems to
address and solve this issue [18].

Author Contributions: P.M.S.: conceptualization, methodology, software, validation, formal analysis,

investigation, writing and editing, visualization. D.S.: supervision; project administration and critical
revision. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Acknowledgments: We acknowledge support by the Open Access Publication Fund of the University
of Duisburg-Essen.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Johansson, R. Vision Zero—Implementing a policy for traffic safety. Safety Sci. 2009, 47, 826–831. [CrossRef]
2. Elbanhawi, M.; Simic, M.; Jazar, R. In the Passenger Seat: Investigating Ride Comfort Measures in Autonomous Cars. IEEE Intell.
Transp. Syst. Mag. 2015, 7, 4–17. [CrossRef]
3. Chen, W.; Xiao, H.; Wang, Q.; Zhao, L.; Zhu, M. Integrated Vehicle Dynamics and Control; John Wiley & Sons Singapore Pte. Ltd.:
Singapore, 2016.
4. Fan, Y.; Dao-Fei, L.; Crolla, D.A. Integrated Vehicle Dynamics Control—State-of-the art review. In Proceedings of the 2008 IEEE
Vehicle Power and Propulsion Conference, Harbin, China, 3–5 September 2008; pp. 1–6.
5. Sieberg, P.M.; Blume, S.; Reicherts, S.; Schramm, D. Nichtlineare modellbasierte prädiktive Regelung der Fahrzeugdynamik in
Bezug auf eine aktive Wankstabilisierung und eine Nickreduzierung. In Forschung im Ingenieurwesen; Springer: Berlin/Heidelberg,
Germany, 2019; Volume 83, pp. 119–127.
6. Ikenaga, S.; Lewis, F.L.; Campos, J.; Davis, L. Active suspension control of ground vehicle based on a full-vehicle model. In
Proceedings of the 2000 American Control Conference, ACC (IEEE Cat. No.00CH36334), Chicago, IL, USA, 28–30 June 2000;
Volume 4016, pp. 4019–4024.
7. Chang, S.; Gordon, T.J. A flexible hierarchical model-based control methodology for vehicle active safety systems. Veh. Syst. Dyn.
2008, 46, 63–75. [CrossRef]
8. Bahr, M.; Reicherts, S.; Sieberg, P.; Morss, L.; Schramm, D. Application of Artificial Neural Networks for Active Roll Control
Based on Actor-Critic Reinforcement Learning. In Proceedings of the Simulation and Modeling Methodologies, Technologies and
Applications, Prague, Czech Republic, 29–31 July 2021; pp. 61–82.
Appl. Sci. 2021, 11, 4687 18 of 18

9. Li, H.-m.; Wang, X.-b.; Song, S.-B.; Li, H. Vehicle Control Strategies Analysis Based on PID and Fuzzy Logic Control. Procedia Eng.
2016, 137, 234–243. [CrossRef]
10. Sands, T. Development of Deterministic Artificial Intelligence for Unmanned Underwater Vehicles (UUV). J. Mar. Sci. Eng. 2020,
8, 578. [CrossRef]
11. Camacho, E.F.; Bordons, C.A. Model Predictive Control; Springer: Berlin, Germany, 1998.
12. Grüne, L.; Pannek, J. Nonlinear Model Predictive Control. In Nonlinear Model Predictive Control: Theory and Algorithms; Grüne, L.,
Pannek, J., Eds.; Springer: Cham, Switzerland, 2017; pp. 45–69. [CrossRef]
13. Beal, C.E.; Gerdes, J.C. Predictive control of vehicle roll dynamics with rear wheel steering. In Proceedings of the 2010 American
Control Conference, Baltimore, MD, USA, 30 June–2 July 2010; pp. 1489–1494.
14. Mehra, R.K.; Amin, J.N.; Hedrick, K.J.; Osorio, C.; Gopalasamy, S. Active suspension using preview information and model
predictive control. In Proceedings of the 1997 IEEE International Conference on Control Applications, Hartford, CT, USA, 5–7
October 1997; pp. 860–865.
15. Canale, M.; Milanese, M.; Novara, C. Semi-Active Suspension Control Using “Fast” Model-Predictive Techniques. IEEE Trans.
Control Syst. Technol. 2006, 14, 1034–1046. [CrossRef]
16. Hrovat, D. Survey of Advanced Suspension Developments and Related Optimal Control Applications. Automatica 1997, 33,
1781–1817. [CrossRef]
17. Giua, A.; Seatzu, C.; Usai, G. Semiactive Suspension Design with an Optimal Gain Switching Target. Veh. Syst. Dyn. 1999, 31,
213–232. [CrossRef]
18. Sieberg, P.M.; Hürten, C.; Schramm, D. Representation of an Integrated Non-Linear Model-Based Predictive Vehicle Dynamics
Control System by a Co-Active Neuro-Fuzzy Inference System. In Proceedings of the 2020 IEEE Intelligent Vehicles Symposium
(IV), Las Vegas, NV, USA, 19 October–13 November 2020; pp. 572–577.
19. Sieberg, P.M.; Blume, S.; Reicherts, S.; Maas, N.; Schramm, D. Hybrid State Estimation—A Contribution towards Reliability
Enhancement of Artificial Neural Network Estimators. IEEE Trans. Intell. Transp. Syst. 2021. [CrossRef]
20. Sieberg, P.M.; Blume, S.; Schramm, D. Side-Slip Angle Estimation by Artificial Neural Networks for Vehicle Dynamics Control
Applications. In Proceedings of the AmE 2021—Automotive meets Electronics 12th GMM-Symposium, Online, 10–11 March
2021.
21. Blume, S.; Sieberg, P.M.; Maas, N.; Schramm, D. Neural Roll Angle Estimation in a Model Predictive Control System. In
Proceedings of the IEEE Intelligent Transportation Systems Conference (ITSC)/ITSC 2019, Auckland, New Zealand, 27–30
October 2019; pp. 1625–1630.
22. Rossiter, J.A. Model-Based Predictive Control—A Practical Approach; CRC Press: Boca Raton, FL, USA, 2004; p. 344.
23. Cromer, A. Stable solutions using the Euler approximation. Am. J. Phys. 1981, 49, 455–459. [CrossRef]
24. Schramm, D.; Hesse, B.; Maas, N.; Unterreiner, M. Vehicle Technology—Technical Foundations of Current and Future Motor Vehicles;
De Gruyter Oldenbourg: Berlin, Germay; Boston, MA, USA, 2020. [CrossRef]
25. Schramm, D.; Hiller, M.; Bardini, R. Vehicle Dynamics: Modeling and Simulation, 2nd ed.; Springer: Berlin/Heidelberg, Germany,
2018. [CrossRef]
26. Butcher, J.C. Numerical Methods for Ordinary Differential Equations, 3rd ed.; John Wiley & Sons, Ltd.: Hoboken, NJ, USA, 2016.
[CrossRef]
27. Pacejka, H.B. Tyre and Vehicle Dynamics, 2nd ed.; Butterworth-Heinemann: Oxford, UK, 2006; p. 1.
28. Waltz, R.A.; Morales, J.L.; Nocedal, J.; Orban, D. An interior algorithm for nonlinear optimization that combines line search and
trust region steps. Math. Program. 2006, 107, 391–408. [CrossRef]
29. Ypma, T.J. Historical Development of the Newton-Raphson Method. SIAM Rev. 1995, 37, 531–551. [CrossRef]
30. Yildirim, E.A.; Wright, S.J. Warm-Start Strategies in Interior-Point Methods for Linear Programming. SIAM J. Optim. 2002, 12,
782–810. [CrossRef]
31. ISO. ISO 3888-1:2018 Passenger Cars—Test Track for a Severe Lane-Change Manoeuvre—Part 1: Double Lane-Change; International
Organisation for Standardization (ISO): Geneva, Switzerland, 2018.
32. ISO. ISO 13674-1:2010 Road Vehicles—Test Method for the Quantification of on-Centre Handling—Part 1: Weave Test; International
Organisation for Standardization (ISO): Geneva, Switzerland, 2010.
33. Sieberg, P.M.; Reicherts, S.; Schramm, D. Nichtlineare modellbasierte prädiktive Regelung zur aktiven Wankstabilisierung von
Personenkraftwagen. In Proceedings of the Vierte IFToMM D-A-CH Konferenz 2018, Lausanne, Switzerland, 15–16 February
2018; pp. 1–8.