energies

Article
A Novel Data-Driven Modeling and Control Design Method for
Autonomous Vehicles
Dániel Fényes 1 , Balázs Németh 2 and Péter Gáspár 2, *

1 Department of Control for Transportation and Vehicle Systems, Budapest University of Technology and
Economics, H-1111 Budapest, Hungary; [email protected]
2 Systems and Control Laboratory, Institute for Computer Science and Control (SZTAKI), H-1111 Budapest,
Hungary; [email protected]
* Correspondence: [email protected]

Abstract: This paper presents a novel modeling method for the control design of autonomous
vehicle systems. The goal of the method is to provide a control-oriented model in a predefined
Linear Parameter Varying (LPV) structure. The scheduling variables of the LPV model through
machine-learning-based methods using a big dataset are selected. Moreover, the LPV model param-
eters through an optimization algorithm are computed, with which accurate fitting on the dataset
is achieved. The proposed method is illustrated on the nonlinear modeling of the lateral vehicle
dynamics. The resulting LPV-based vehicle model is used for the control design of path following
functionality of autonomous vehicles. The effectiveness of the modeling and control design meth-
ods through comprehensive simulation examples based on a high-fidelity simulation software are
illustrated.

Keywords: LPV control design; machine learning in modeling; vehicle dynamics






1. Introduction and Motivation
Citation: Fényes, D.; Németh, B.;
Nowadays, one of the major challenges for the automotive industry is the development
Gáspár, P. A Novel Data-Driven
of the fully autonomous vehicles, which is represented by level 5 of SAE J3016. This
Modeling and Control Design Method
challenge involves several disciplines (e.g., sensing, controlling, decision making), which
for Autonomous Vehicles. Energies
must work together to develop safe and reliable solutions for autonomous vehicles. One of
2021, 14, 517. https://doi.org/
the most significant tasks is the control design, because in the dynamics of the vehicle, it is
10.3390/en14020517
highly nonlinear in its entire operation range. The nonlinear behavior becomes especially
Received: 14 December 2020
important at dangerous maneuvers (e.g., low road adhesion, high speed etc.), in which the
Accepted: 15 January 2021
vehicle is close to its physical limits.
Published: 19 January 2021 Controlling of nonlinear system is still a challenging task, which draws attention
from both engineers and researchers. In the past decades, several algorithms and control
Publisher’s Note: MDPI stays neu- methods have been developed to deal with this problem. The developed approaches in two
tral with regard to jurisdictional clai- main categories can be classified: classical methods and machine learning-based solutions.
ms in published maps and institutio- The classical methods includes the model-based control analysis and synthesis tech-
nal affiliations. niques, i.e., linear robust and optimal control methods (H2 , H∞ ), parameter-varying meth-
ods, e.g., Linear Parameter Varying (LPV) and Nonlinear Parameter Varying (NLPV) [1,2],
Model Predictive Control (MPC) approaches [3] or polynomial methods [4,5]. One of the
main sources of the nonlinearities is the tire-road contact, it stability is analyzed in [6]
Copyright: © 2021 by the authors. Li-
and a Lyapunov control method for this problem is presented in [7]. The uncertainties
censee MDPI, Basel, Switzerland.
and the nonlinearities of the vehicle may have external sources, which can be handled
This article is an open access article
by an adaptive robust controller, see [8]. The steering system and its actuator also have
distributed under the terms and con-
ditions of the Creative Commons At-
nonlinearities, which can be controlled by an LPV-based controller [9]. The identification of
tribution (CC BY) license (https://
the uncertainties and nonlinearities is also a challenging task. The authors of [10] present a
creativecommons.org/licenses/by/ grey-box LPV identification method for vehicles using side-slip angle estimation. An LPV
4.0/). framework-based uncertainty identification is proposed in [11] using experimental data.

Energies 2021, 14, 517. https://doi.org/10.3390/en14020517 https://www.mdpi.com/journal/energies

Energies 2021, 14, 517 2 of 16

These algorithms handle nonlinearities and unmodelled dynamics in the form of

uncertainties and thus, they provide conservative solutions. Nevertheless, the result of the
design is a control loop with guarantees on the stability and on the performances.
The machine-learning-based solutions include various methods, such as deep learning
through the training of neural networks [12], Support Vector Machine (SVM), or decision
logic algorithms [13–15]. The main advantage of these algorithms is that they can learn
from data, thus their models inherently contain the nonlinear behavior of the control
plant. Thus, in several control applications, better performances comparing to the classical
approaches might be provided. However, some of these methods, especially the neural
networks, have the drawback that it is difficult to find systematic methods to prove the
stability and performances of the closed-loop system. Since autonomous vehicles are
safety-critic systems, it is recommended to reformulate the learning problem.
There are other solutions that deal with the identification problem of the dynamical
systems. For example, in [16,17] SVM-LS-based LPV system identification methods are
presented. Another data-driven identification solution can be found in [18], which can ease
the tuning of the hyperparameters of the model identification process. A non-parametric
and probabilistic approach is proposed by [19] for identifying nonlinear system considering
uncertain noises on the measured signals. particle Bernstein polynomials-based regression
method is presented in [20], which is suitable for multivariate regression problems.
In the literature, some solutions can be found, which can provide methods for the
reformulation of the learning features in the control problem. For example, [21] proposes a
MPC-based control solution, in which the terminal cost and set are determined through
an iterative process. In this idea, the model of the system is based on physical principles
with its limitations. Another solution is the Model Free Control (MFC), which is proposed
in [22,23]. The MFC method does not require a model of the controlled system, but it uses
a local model, in which the model of each time step is updated.
The goal of this paper is to reformulate the design of vehicle path tracking functionality
as a modeling problem with learning features and a control design problem using a model-
based approach. The main contribution of the paper is a novel identification method, by
which an LPV model of the nonlinear system can be determined. Moreover, the selection
of the scheduling parameters of the LPV system is determined by a machine-learning-
based pace regression algorithm in the modeling process of the system. In this way, the
performances of the LPV controller are enhanced, while the stability of the closed-loop
system is also preserved. The efficiency and the operation of the proposed control algorithm
is validated by a complex test scenario, which is performed in the high-fidelity simulation
software, CarSim.
The process of the proposed method for achieving an enhanced vehicle control system
is presented in Figure 1. The process consists of three main layers with several tasks, i.e.,
Simulation environment, Modeling error computation using ML algorithm and Model identification
and control design. Layer Simulation environment consists of two tasks, namely, performing
vehicle dynamic simulations for data generation and data acquisition. The vehicle dynamic
simulations are performed on the high-fidelity simulation environment CarSim. The
layer Modeling error computation using ML algorithm consists of several tasks. It contains
the selection and categorization of the scheduling parameters using the pace regression
algorithm. It provides the categorized datasets for the identification process. The Model
identification and control design uses the categorized datasets for computing the models for
each grid points of the LPV system. In this layer, the LPV-based control design is also
presented, which guarantees the trajectory tracking of the vehicle.
In the paper, each layer and task are presented as follows. Section 2 presents the
vehicle model parameter tuning method for achieving control-oriented LPV model. The
control design based on the robust LPV synthesis method is proposed in Section 4. The
effectiveness of the resulting control system through comparative simulation scenarios is
examined in Section 5. Finally, the achievements of the paper are concluded in Section 6.
Energies 2021, 14, 517 3 of 16

Modeling error
Computation of
Simulation computation using
nominal outputs
Environment ML algorithm
(CarSim)

Categorization
Data Error Machine learning
and selection of
acquisition functions algorithm
scheduling par.
measured attributes

Vehicle controller
Control design
Optimization of
simulator vehicle model

Model identification and control design

Figure 1. Methodological process for modeling, control design, and evaluation.

2. Formulation of the Control-Oriented LPV Model Using Data-Driven Approach

In this section, the data acquisition and the LPV-based data-driven model parameter
tuning process are presented. Firstly, the collection of the dataset with numerous elements is
detailed, which is provided by the vehicle dynamics simulation software, CarSim. Secondly,
the prepocess of the collected data is explained, which includes the selection and the
scaling of the data. Finally, the data-driven LPV-based model parameter tuning process
is presented, in which the scheduling parameters and the model parameters are selected
through a machine-learning-based pace regression algorithm.

2.1. Acquisition of Data from Simulations

Since all of the machine learning methods require a large amount of data to provide
satisfying results, the first the of the algorithm is the data acquisition. In this paper, the
high-fidelity simulation software, CarSim is used to generate the datasets for the machine
learning algorithm in the control-oriented model formulation process. Since an appropriate
dataset must cover a wide range of the vehicle operation, several simulations have been
performed on the simulation environment using different parameter sets. These parameter
sets includes the longitudinal velocity profile, the maneuvers, which must be performed
by the vehicle. The test scenarios included several different routes with sharp bends
(e.g., Melbourne Formula (1) circuit), whose adhesion coefficient has also been modified
µ ∈ [0.4 . . . 1] during the simulations. The dataset also includes simulations, in which the
vehicle loses its stability, in order to cover most of the possible situations that the vehicle
can encounter.
During the simulations, the following variables have been measured and collected:
1. longitudinal velocity (v x )
2. angular velocity of the wheels (ωx,y ), x ∈ { f ront, rear }, y ∈ {le f t, right}
3. steering angle (δ)
4. yaw-rate (ψ̇)
5. accelerations (a x , ay )
6. lateral velocity (vy ).
7. side-slips of the wheels (α x ), x ∈ { f ront, rear }.
Note that all of the collected signals are available from the onboard system of the
vehicle except the lateral velocity (vy ) and the side-slip angles. These signals are only used
in the model formulation process, but during the operation of the proposed control method
they are not required. The sampling time of the variables has been set to Ts = 0.01 s. In this
way, a large dataset has been created, which consisted of more than 10 million instances.
Energies 2021, 14, 517 4 of 16

2.2. Categorization of Instances

The goal of the parameter optimization process is to get a set of models which are
stable, as a first step the instances in the dataset must be categorized by their stability.
The stability of the instances have a high impact on the performances of the optimization
process; thus, the unstable instances must be removed from the dataset [24]. The separation
of the stable/unstable instances using only the measured values is not straightforward
task. There is no general solution for determining the stability of a measurement. However,
for the automotive case, a criterion is proposed in [25], which can be used to separate the
instances according to the stability of the vehicle:

|1 + α1 |
−ε < l1 ψ̇
− 1 ≤ ε, (1)
|1 + δ − β − vx |

where e is a experimentally defined parameter.

This criterion is based on the linear bicycle model, which will be detailed in the
followings (2). Briefly, it compares the measured side-slips angle of the front wheel α1 , to
l ψ̇
the linear model, in which that angle is computed as α1 = δ − β − v1 x . The inequality (1)
describes that the deviation of the measured instance from the linear region, which can
be used to determine the stability of the vehicle. Using this criterion, the dataset can be
divided into two categories: Rst represents the set of the instances, where the motion of the
vehicle is approximated as stable. Similarly, Rust denotes the set consisting of the instances
with the approximation of unstable vehicle motion.
After separating the instances, error functions are computed, which reflects on the
nonlinear behavior of the vehicle. Basically, the error function describes the deviation
between the nominal model and the measured variables. The nominal model, which is
used in this paper, is the two-wheeled bicycle model. The model consists of the following
three main equations, ref. [26]:

I ψ̈ = C1 α1 l1 − C2 α2 l2 (2a)
may = C1 α1 + C2 α2 (2b)
ay = v̇y + v x ψ̇ (2c)

where l1 , l2 are the distances between the CoG and the front, rear axes of the vehicle, C1 , C2
are the reduced cornering stiffness of the wheels, ψ̇ denotes the yaw-rate, β is the slid-slip,
m is the mass of the car, I represents the yaw-inertia, α1 , α2 are the side-slip of the wheels
and the longitudinal and lateral velocities are denoted by v x , vy . The lateral vehicle model
can be transformed into a state-space representation

ẋ = Ax + Bu, (3)

whose state-vector consists of x = [ψ̇ vy ] T and its control input is the steering angle u = δ.
Furthermore, the measured variables have a sampling time of Ts = 0.01 s, therefore the
model is sampled by the same sampling time. Using the measured input signal δ, the
outputs of the discrete system are computed for each measurement point.
The labeling of the collected data is based on the deviation of the measured signals
from the signals of the nominal system. In this process, the yaw-rate and the lateral
velocities are involved, which are the independent states of the physical system. The
labeling is based on the relative errors of the signals in time ti , such as

∆ψ̇ = ψ̇m (ti ) − ψ̇n (ti ) (4a)

∆vy = vy,m (ti ) − vy,n (ti ) (4b)

where ψ̇m and vy,m denote the measured outputs while ψ̇n and vy,n are the outputs of the
nominal system.
Energies 2021, 14, 517 5 of 16

Figure 2 shows the function of ∆ψ̇ − ∆vy computed from the collected dataset.

Figure 2. Illustration of the error functions and their resolution.

Since neither ∆ψ̇ nor ∆vy are easy to calculate during the operation of the vehicle, a
method must be found that is able to appropriately approximate the error signals using
only the available measurements. In this paper the pace-regression algorithm is used for
this purpose, which is briefly introduced in the following subsection.

3. Parameters Optimization and Determination of Scheduling Parameters

3.1. Fundamentals of the Applied Machine-Learning-Based Method
The main goal of the algorithm is to compute (or approximate) the selected output
signal y by using another measured attributes, which are written in a matrix X. The
algorithm tries to find the parameter vector ξ ∗ , which is the parameter vector of the true
model and the output can be computed as

y = Xζ ∗ + e (5)

where e denotes the noises, which are computed from N (0, σ2 ). σ2 must be determined or
estimated (σ̂2 ). The fitted linear model is denoted by M(ζ ), which has a parameter vector
ζ. The goal of the optimization process is to find the best model from the model space
M = {M(ζ ) : ζ ∈ Rk }, which has the lowest error on the whole dataset. There are several
algorithms, which are able to solve this problem shrinkage, OLS subset selection, CIC, RIC,
methods, etc., see [27]. Sometimes, solving this problem may be difficult; therefore, these
algorithms can eliminate the redundant and irrelevant variables from the dataset before
starting the optimization. The produced models are evaluated by their euclidean distance:

||yM(ζ ∗ ) − yM(ζ ) ||2

D(M(ζ ∗ ), M(ζ )) = (6)
σ2

where || · || denotes the L2 norm and σ2 can be replaced by its estimated value σ̂2 . The final
task is to determine a model which minimizes this expression.

D(M∗ , M) = min! (7)

Energies 2021, 14, 517 6 of 16

The formulation of the appropriate model may be difficult when all of the attributes
are used. Therefore, it is recommended to create subsets using reduced number of variables
to get better results. Suppose that the dataset has k variables, this means that k + 1 can
be created including the null model (j = 0) and the full model j = k. Then the parameter
vector for each model (M j ) can be computed as:

0 −1 0
ζ̂ M j = ( XM j XM j ) XM j y (8)

where XM j is the n × j design matrix and let PM j = XM j ( XM 0 X −1 0

j M j ) XM j be an or-
thogonal projection matrix from the original space (k) onto the reduced space (j). Finally,
ŷM j = PM j y is the estimate of y∗M = PM j y∗ . In order to find the best fitting model,
j
numerous models must be evaluated. Each model contains different set of variables. The
number of all the possible combinations of the variables is 2k . At increased k the computa-
tion and the evaluation of 2k models are not feasible. Therefore, the algorithm reduces the
number of the possible cases from 2k to k + 1. Firstly, the predefined order of the variables is
computed, this order reflects on the correlation between the variables and output. The first
variable is the most correlated to the output and the last variable is the least correlated. The
ordering can be done by using several algorithms such as [28]. After the ordering process,
the algorithm has to evaluate only k + 1 models starting from the full model (including
all variables) to the zero model (no variables). Finally, the best model can be easily chosen
from the k + 1 evaluated models.
The pace regression algorithm is used to compute the scheduling variables of the LPV
system ∆ψ̇, ∆vy from the dataset, which contains the measured attributes. The result of the
pace-regression algorithm is a model which can be used to approximate ∆ψ̇ˆ , ∆vˆy .

3.2. Parameter Selection of the Control-Oriented Model

The goal of the identification process is to determine the parameters of the model (10)
for each segment. The structure of the state-space representation is determined in such a
way to preserve the original structure of the physical model (2). In this case, the lateral
model can be formed as:

ẋd = Ad (ρ) xd + Bd ud (ρ), (9)

where
   
a (ρ) a12 (ρ) b1 (ρ)
Ad (ρ) = 11 , Bd (ρ) = , (10)
a21 (ρ) a22 (ρ) b2 (ρ)

and a11 (ρ), a12 (ρ), a21 (ρ), a22 (ρ) and b1 (ρ), b2 (ρ) are parameters and the state-vector of the
system is xd = [ψ̇i vy ], the control input is ud = δ. The scheduling parameters are written
into a vector form: ρ = [∆ψ̇, ∆vy , v x ]. Since the parameter optimization process is
especially difficult for continuous variables, the optimization is performed for a finite
number of operating points. Each operating point is represented by constant vector ρ,
in which the scheduling variables are fixed at constant values. The resolution of the
scheduling variables is a crucial aspect of the parameter identification process. In order
to cover the nonlinear dynamics of the vehicle precisely, the resolution must be as high
as possible. However, the high resolution may make the computation of the system and
the controller difficult. Therefore, a balance must be found between them. In this case,
equidistant resolution is used: Γψ̇ , Γvy as shown in Figure 3. The last scheduling variable
v x is also ordered into a finite number of groups. nvx , n∆vy and n∆ψ̇ represent the numbers
of groups of the scheduling parameters.
Energies 2021, 14, 517 7 of 16

Figure 3. Resolution of the scheduling variables.

The main goal if the identification process is to determine the parameters a11 (ρi ),
a12 (ρi ), a21 (ρi ), a22 (ρi ) and b1 (ρi ), b2 (ρi ) for each segment, where ρi denotes a specific
operating point of the system with fixed ranges of the scheduling parameters. It can be
written into an optimization problem:

N
2
min
a11 (ρi ),a12 (ρi ),a21 (ρi ),a22 (ρ ),b (ρ ),b (ρ )
∑ xm,ρi (t j ) − x (t j ) (11)
i 1 i 2 i j =0

where xm,ρi (tji ) denotes the instances of the dataset, which belong to the operating range
defined by ρi . N is the number of the samples within given operating range. x (t j ) is the
output of the nominal system. During the solution of (11) the systems can be computed
independently for fixed ρi values in the grids. Thus, the parameter-dependent quadratic op-
timization problem leads to a least-squares problem [29,30]. The result of the optimization
is a set of polytopic systems, which represents the LPV description of the vehicle model.

3.3. Evaluation of the Data-Driven LPV Models

In the following, a test case is presented to show the efficiency of the proposed
parameter optimization method. The outputs of the optimized system are compared to
the outputs of a nominal model presented in Section 2. The parameters of the nominal
model are given by the simulation software (CarSim) such as mass, inertia, geometrical
parameters, etc. Specific values can be found in Table 1. In the simulation, the vehicle
is controlled by the in-built driver model of CarSim, and it is driven along a segment of
Michigan Waterford Hill Track. In Figure 4, we show the results of the control-oriented
LPV system for a simulation-based test case. The dashed red lines represent the measured
outputs (ψ̇, vy ) of the nonlinear vehicle model from CarSim, dotted yellow lines are the
outputs of the nominal LPV models (see (2)) and blue lines illustrate the outputs of the
identified system. As it can be seen, the error between the measured and the computed
outputs are smaller when the identified model is used.
Energies 2021, 14, 517 8 of 16

0.6 0.5
Identified LPV model
0.5 Measured signal
0.4 Identified LPV model
Nominal LPV model
Measured signal
0.4 Nominal LPV model
0.3

0.3
0.2

0.2

vy (m/s)
0.1
0.1
0
0

-0.1
-0.1

-0.2
-0.2

-0.3 -0.3

-0.4 -0.4
0 50 100 150 200 250 0 50 100 150 200 250
Time (s) Time (s)

(a) yaw-rate outputs (b) lateral velocity outputs

0.3 0.3

0.2 0.2

0.1 0.1

0 0

-0.1 -0.1

-0.2 -0.2

-0.3 -0.3
-0.15 -0.1 -0.05 0 0.05 0.1 0 50 100 150 200 250
Time (s)

(c) ∆vy vs ∆ψ̇ (d) scheduling parameters with respect to time

Figure 4. Evaluation of the optimized model.

The difference is significant in case of vy . When the nominal model is used, the
averaged error is ≈0.07 m/s, in the case of the identified system it reduces to ≈0.017 m/s.
The scheduling parameters are shown in Figure 4c,d. A significant range of scheduling
parameters can be seen, which means that the identified system works well at different
operating points. It can be said that the identified system together with the selected
scheduling parameters fits better to the nonlinear model than the nominal model.

4. Path Following LPV Control Design Using the Data-Driven Model

In this section , the LPV-based lateral control design is presented. However, in some
situations, the intervention is very limited using only the steering angle. It must be noted
that, at specific operating points, the performance of the nominal intervention (steering
angle) is limited due to the nonlinear dynamics of the vehicle. Therefore, the model
is extended with the differential drive has a second input signal, which will be used
to compensate the degraded performances of the steering system. This means that the
model (2) is modified as:

I ψ̈ = C1 α1 l1 − C2 α2 l2 + Md (12)

where Md denotes the differential torque. Moreover, the original model contains only
two states: yaw-rate and the lateral velocity in order to simplify the optimization task by
minimizing the number of the parameters. However, the lateral position tracking is also
Energies 2021, 14, 517 9 of 16

an important requirement for the autonomous vehicle, which can be computed from the
existing states. Therefore, the state-space model is extended with at the lateral position y,
Rt
which can be computed as y(t) = T =1 vy ( T )dT at tth s.
The identified system description is augmented as:

ẋ = A(ρ) x + B(ρ)u (13)

and
1
   
a11 (ρ) a12 (ρ) 0 b1 (ρ) Iz
A(ρ) =  a21 (ρ) a22 (ρ) 0, B(ρ) = b2 (ρ) 0 ,
0 1 0 0 0
 T
where the state vector of the system is x = ψ̇ vy y , the input is u = [δ Md ].
The main goal of the control design is to guarantee the trajectory tracking of the
vehicle, which can be achieved by describing performances for controller:
• Minimization of the lateral error: As mentioned, the goal of the control design is to
guarantee the trajectory tracking of the vehicle thus the error between the measured
and the reference lateral positions must be minimized:

z2 = yre f − y, |z2 | → min, (14)

where yre f is the reference lateral position computed from the track.
• Minimization of the yaw-rate error: In order to achieve smooth trajectory tracking, a
reference yaw-rate is also prescribed, which also must be tracked by the vehicle:

z1 = ψ̇re f − ψ̇, |z1 | → min, (15)

where ψ̇re f is the reference yaw-rate signal, computed from the curvature of the road,
see [26].
• Minimization of the interventions: Due to the energy consumption, the interventions
also must be minimized during the operation of the vehicle:

z3 = δ, |z3 | → min. (16)

z 4 = Md , |z4 | → min. (17)
 T
The performances are compressed into a performance-vectorz = z1 z2 z3 z4 , which
leads to the following performance equation

z = C1 x + D11 r + D12 u, (18)

where C1 , D11 , D12 are matrices and r contains the reference signals yre f and ψ̇re f .
The basis of the control design is the presented data-driven LPV model (13), its
measurement and performance equations are:

ẋ = Ax + Bu, (19a)
z = C1 x + D11 r + D12 u, (19b)
yK = C2 x, (19c)
 
where (19c) represents the measurement of yK = y ψ̇ .
In order to guarantee the predefined performances ((14)–(16)), the system is aug-
mented with several weighting functions. The augmented system is illustrated in Figure 5.
The weighting functions Wz,1 , Wz,2 aim to guarantee the tracking performances, while
Energies 2021, 14, 517 10 of 16

Wz,3 , Wz,3 aim to minimize the interventions. Since the parallel minimization of the perfor-
mances and the interventions is contradictory, another goal of the weighting functions are
to describe a balance between them.
Wre f ,1 and Wre f ,2 are used to scale the references signals (ψ̇re f , yre f ). Moreover, the role
of the functions Ww,1 , Ww,2 is to attenuate the noises on the measured signals.

Figure 5. Augmented plant.

The goal of the control design is to find a K (ρ) controller, which can guarantee the
predefined performances and by which the closed-loop system is quadratically stable and
its induced L2 norm is smaller than a γ value. The optimization task can be formed as:

k z k2
inf sup sup , (20)
K (ρ) ρ∈ Fρ kwk 6=0,w∈L
2
k w k2
2

where Fρ bounds the scheduling variables. The yielded controller K (ρ) is formed as

ẋK = AK (ρ) xK + BK (ρ)yK , (21a)

u = CK (ρ) xK + DK (ρ)yK , (21b)

where AK (ρ), BK (ρ), CK (ρ), DK (ρ) are variable-dependent matrices. The interconnected
system is illustrated in Figure 6 showing all of the main components of the proposed
control system.
Energies 2021, 14, 517 11 of 16

Measurements

Figure 6. Structure of control system.

5. Simulation Results
In the followings, a complex test scenario is presented to show the efficiency and
the operation of the proposed control algorithm in the high-fidelity vehicle dynamics
simulation software, CarSim. In the simulations, the vehicle is driven along the Michigan
Waterford hill track twice. In the first run, the vehicle is controlled by the proposed
data-driven LPV controller, while in the second turn, the car is driven by a nominal LPV
controller. Note that the nominal LPV controller is tuned similarly to the proposed LPV
controller using the same number of weighting functions. However, the weights of the
nominal LPV controller are optimized to the nominal model presented in Section 2.
The parameters of a car (e.g., mass, inertia, geometric attributes) are listed in Table 1.

Table 1. Parameters of the used D-class vehicle.

Parameter Notion Value Unit

Mass of the car m 1690 kg
Yaw-inertia J 4192 kg/m2
Location of front axis from COG l1 1.11 m
Cornering stiffness of front wheels C1 155,160 N/rad
Location of rear axis from COG l2 1.66 m
Cornering stiffness of rear wheels C2 114,659 N/rad
Front drag area of the car A 1.8 m2
Height of COG h 0.56 m
Type of front suspensions - Independent -
Mass of front suspensions ms, f 85 kg
Type of rear suspensions - Independent -
Mass of rear suspensions ms,r 85 kg

As presented in Section 4, the LPV controller contains several weighting functions,

which aim to guarantee the predefined performances. The weighting functions of the
reference signals are designed in such a way to ensure smooth trajectory tracking, while
guaranteeing small tracking error:

0.1 0.01
Wre f ,1 = , Wre f ,2 = . (22)
15 s + 1 20 s + 1
Energies 2021, 14, 517 12 of 16

The goal of the scaling functions Wz,1 , Wz,2 are to guarantee the trajectory tracking of the
vehicle, they are designed by considering the maximal allowable errors, see:

5 15
Wz,1 = , Wz,2 = . (23)
0.3 s + 1 0.5 s + 1
The functions Wz,3 and Wz,2 aim to ensure the balance between the interventions and
the tracking performances. As it can be seen, the weighting function belonging to the
differential drive depends on the scheduling parameters ∆ψ̇ˆ and ∆v̂y . It is tuned to
compensate for the degraded performances of the steering system at the highly nonlinear
region.

1s+1 1 1s+1
Wz,3 = 0.1 , Wz,4 = . (24)
2s+1 ˆ
(10∆ψ̇∆v̂y ) 4 s + 1
2

Finally, the weighting functions of the measurements are presented. They are used to
attenuate the noises on the measured signals, which may have a negative effect on the
performances.

Ww,1 = 0.02, Ww,2 = 0.05. (25)

In the followings, the results of the test scenarios are presented. As mentioned, the
vehicle has been driven along the track twice. The track and the trajectories of the vehicle
are illustrated in Figure 7. As it can be seen, when the vehicle is controlled by the nominal
LPV controller, it leaves the road at a sharp bend. Whilst, in the second case, when the car
is controlled by the proposed data-driven LPV controller, it follows the road throughout
the whole track.

150 -40

Proposed LPV controller Proposed LPV controller

Nominal LPV controller Nominal LPV controller
100 -60

50 -80

0 -100
Y (m)

Y (m)

-50 -120

-100 -140

-150 -160

-200 -180
-200 -100 0 100 200 300 400 -100 -50 0 50 100
X (m) X (m)

(a) Whole track (b) Highlighted section

Figure 7. Positions of the vehicles during the simulations.

In order to show the operation of the proposed control method in a wide range,
the longitudinal velocity of the vehicle varies between v x ∈ <40, 73> km/h, as shown in
Figure 8a. Figure 8 demonstrates the lateral acceleration of the vehicle. As it can be seen,
the maximum of the lateral acceleration is ≈10 m/s2 , which means that the vehicle is close
to its physical limits. This situation cannot be handled by the nominal controller; thus, the
vehicle leaves the road.
Energies 2021, 14, 517 13 of 16

70 10

8
65

6
60
4

55

a y (m/s2)
vx (km/h)
2

50 0

-2
45
-4

40
-6

35 -8
0 50 100 150 0 50 100 150
Time (s) Time (s)

(a) Velocity profile (b) Lateral acceleration

Figure 8. Velocity profile and lateral acceleration of the vehicle.

Figure 9 shows the scheduling parameters and the yaw-rate tracking of the vehicle. As
Figure 9a demonstrates, the scheduling parameters cover almost its whole range, as shown
in Figure 4. The yaw-rate tracking of the vehicle can be seen in Figure 9b. The tracking
is accurate the maximum error is only ≈0.02 rad/s. The yaw-rate signal varies in a wide
range, its maximum is close to 0.6 rad/s, which is also close to the physical limits of the
vehicle.

0.8 0.6
Vehicle
Reference
0.5
0.6
0.4

0.4 0.3

0.2
0.2

0.1

0
0

-0.2 -0.1

-0.2
-0.4
-0.3

-0.6 -0.4
-0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0 50 100 150
Time (s)

(a) Scheduling parameters (b) Tracking of yaw-rate

Figure 9. Scheduling parameters and tracking of yaw-rate.

Finally, the interventions are illustrated in Figure 10. Figure 10a show the differential
torque computed by the proposed control algorithm. As it can be seen, its maximal value
is ≈600 Nm, which is provided at the sharp bend to compensate the steering angle. The
steering angle is depicted in the right figure. It varies between δ ∈ <−0.04, 0.03> rad, which
is a reasonable range for the presented velocity profile.
Energies 2021, 14, 517 14 of 16

800 0.03

0.02
600

0.01
400

0
200
Md (Nm)

(rad)
-0.01

0
-0.02

-200
-0.03

-400
-0.04

-600 -0.05
0 50 100 150 0 50 100 150
Time (s) Time (s)

(a) Differential torque (b) Steering angle

Figure 10. Interventions of the vehicles.

6. Conclusions
In the paper, a novel data-driven identification and control method has been proposed
for autonomous vehicles. The identification process was based on the LPV framework,
in which the scheduling parameters have been computed by a machine learning algo-
rithm (Pace-regression). The efficiency of the identification process has been demonstrated
through an example, and the identified LPV model was the basis of the lateral control
design. The operation and the efficiency of the proposed control system has been demon-
strated through a complex simulation example performed in the high-fidelity simulation
software, CarSim. The simulation example has shown that the proposed control algorithm
provided better performances than a nominal one, especially in extreme maneuvers. The
scheduling parameters of the system covered a wide range, which meant that the the
proposed control algorithm was able to work in the whole range of the vehicle’s operation
range. The main drawback of the proposed algorithm is that it requires a reasonably large
dataset in order to provide the appropriate model for the lateral control design. More-
over, the weighting functions of the controller must be tuned in case of the identified
systems, since the parameters of the system can significantly differ from the nominal model.
However, as shown in the simulation, the identified system-based controller provides
remarkably better performances than the nominal controller.

Author Contributions: Conceptualization, algorithms, software, D.F.; methodology, D.F. and B.N.;
supervision, P.G. All authors have read and agreed to the published version of the manuscript.
Funding: The research reported in this paper and carried out at BME has been supported by the
National Research Development and Innovation Fund (TKP2020 IES, Grant No. BME-IE-MIFM)
based on the charter of bolster issued by the National Research Development and Innovation Office
under the auspices of the Ministry for Innovation and Technology. The research presented in this
paper, carried out by Institute for Computer Science and Control was supported by the Ministry for
Innovation and Technology and the National Research, Development and Innovation Office within
the framework of the National Lab for Autonomous Systems. The work of D. Fényes was supported
by the ÚNKP-20-3 New National Excellence Program of the Ministry for Innovation and Technology
from the source of the National Research, Development and Innovation Fund.The work of B. Németh
was partially supported by the János Bolyai Research Scholarship of the Hungarian Academy of
Sciences and the ÚNKP-20-5 New National Excellence Program of the Ministry for Innovation and
Technology from the source of the National Research, Development and Innovation Fund.
Conflicts of Interest: The authors declare no conflict of interest.
Energies 2021, 14, 517 15 of 16

References
1. Gáspár, P.; Szabó, Z.; Bokor, J.; Németh, B. Robust Control Design for Active Driver Assistance Systems; Springer: Berlin/Heidelberg,
Germany, 2017.
2. Pham, T.P.; Sename, O.; Dugard, L. Real-time Damper Force Estimation of Vehicle Electrorheological Suspension: A NonLinear
Parameter Varying Approach. Part of special issue: 3rd IFAC Workshop on Linear Parameter Varying Systems LPVS 2019:
Eindhoven, Netherlands, 4–6 November 2019. IFAC-PapersOnLine 2019, 52, 94–99. [CrossRef]
3. Rosolia, U.; Borrelli, F. Learning How to Autonomously Race a Car: A Predictive Control Approach. IEEE Trans. Control. Syst.
Technol. 2020, 28, 2713–2719. [CrossRef]
4. Nemeth, B.; Gaspar, P.; Peni, T. Nonlinear analysis of vehicle control actuations based on controlled invariant sets. Int. J. Appl.
Math. Comput. Sci. 2016, 26, 31–43. [CrossRef]
5. Ribeiro, A.M.; Fioravanti, A.R.; Moutinho, A.; de Paiva, E.C. Nonlinear state-feedback design for vehicle lateral control using
sum-of-squares programming. Veh. Syst. Dyn. 2020, 1–27. [CrossRef]
6. Yang, G.; Zhao, Y. Motion Stability Analysis of Vehicle with Four Wheel Steering System Considering Tire Nonlinearity.
In Proceedings of the 2010 3rd International Congress on Image and Signal Processing (CISP2010), Yantai, China, 16–18 October
2010; pp. 3433–3437.
7. Sadri, S.; Wu, Q. Lateral Stability Analysis of On-road Vehicles Using Lyapunov’s Direct Method. In Proceedings of the 2012
Intelligent Vehicles Symposium, Alcala de Henares, Spain, 3–7 June 2012; pp. 821–826.
8. Huijun, G.; Weichao, S.; Shen, Y.; Okyay, K. Stability Control for Lateral Vehicle Motion with Uncertain Parameters and External
Nonlinearities. In Proceesings of the 32nd Chinese Control Conference, Xi’an, China, 26–28 July 2013.
9. Corno, M.; Panzani, G.; Roselli, F.; Giorelli, M.; Azzolini, D.; Savaresi, S.M. An LPV Approach to Autonomous Vehicle Path
Tracking in the Presence of Steering Actuation Nonlinearities. IEEE Trans. Control. Syst. Technol. 2020, 24, 956–970. [CrossRef]
10. Gaspar, P.; Szabo, Z.; Bokor, J. A grey-box identification of an LPV vehicle model with side slip angle estimation. In Proceedings
of the 2007 American Control Conference, Zurich, Switzerland, 4–7 September 2007.
11. Rodonyi, G.; Bokor, J. Uncertainty Identification for a Nominal LPV Vehicle Model Based on Experimental Data. In Proceedings
of the 44th IEEE Conference on Decision and Control, and the European Control Conference 2005, Seville, Spain, 15 December
2005; pp. 2682–2687.
12. Kuutti, S.; Bowden, R.; Jin, Y.; Barber, P.; Fallah, S. A Survey of Deep Learning Applications to Autonomous Vehicle Control.
IEEE Trans. Intell. Transp. Syst. 2020, 1–22. [CrossRef]
13. Hubschneider, C.; Bauer, A.; Doll, J.; Weber, M.; Klemm, S.; Kuhnt, F.; Zollner, J.M. Integrating end-to-end learned steering into
probabilistic autonomous driving. In Proceedings of the 2017 IEEE 20th International Conference on Intelligent Transportation
Systems (ITSC), Yokohama, Japan, 16–19 October 2017; pp. 1–7.
14. Rausch, V.; Hansen, A.; Solowjow, E.; Liu, C.; Kreuzer, E.; Hedrick, J.K. Learning a deep neural net policy for end-to-end control
of autonomous vehicles. In Proceedings of the 2017 American Control Conference (ACC), Seattle, WA, USA, 24–26 May 2017;
pp. 4914–4919.
15. Pomerleau, D. Knowledge-Based Training of Artificial Neural Networks for Autonomous Robot Driving. Robot. Learn. 1993,
233, 13–43.
16. Cavanini, L.; Ferracuti, F.; Longhi, S.; Monteriu, A. LS-SVM for LPV-ARX Identification: Efficient Online Update by Low-Rank
Matrix Approximation. In Proceedings of the 2020 International Conference on Unmanned Aircraft Systems (ICUAS), Athens,
Greece, 1–4 September 2020; pp. 1590–1595.
17. Romano, R.A.; dos Santos, P.L.; Pait, F.; Perdicoulis, T.; Ramos, J.A. Machine learning barycenter approach to identifying
LPV state-space models. In Proceedings of the 2016 American Control Conference (ACC), Boston, MA, USA, 6–8 July 2016;
pp. 6351–6356. [CrossRef]
18. Bao, Y.; Velni, J.M. Data-Driven Linear Parameter-Varying Model Identification Using Transfer Learning. IEEE Control. Syst. Lett.
2021, 5, 1579–1584. [CrossRef]
19. Abdufattokhov, S.; Muhiddinov, B. Stochastic Approach for System Identification using Machine Learning. In Proceedings of
the 2019 Dynamics of Systems, Mechanisms and Machines (Dynamics), Omsk, Russia, 5–7 November 2019; pp. 1–4.
20. Biagetti, G.; Crippa, P.; Falaschetti, L.; Turchetti, C. Machine learning regression based on particle bernstein polynomials for
nonlinear system identification. In Proceedings of the 2017 IEEE 27th International Workshop on Machine Learning for Signal
Processing (MLSP), Tokyo, Japan, 25–28 September 2017; pp. 1–6. [CrossRef]
21. Rosolia, U.; Borrelli, F. Learning Model Predictive Control for Iterative Tasks. A Data-Driven Control Framework. IEEE Trans.
Autom. Control. 2018, 63, 1883–1896. [CrossRef]
22. Fliess, M.; Join, C. Model-free control. Int. J. Control. 2013, 86, 2228–2252. [CrossRef]
23. Formentin, S.; De Filippi, P.; Corno, M.; Tanelli, M.; Savaresi, S.M. Data-Driven Design of Braking Control Systems. IEEE Trans.
Control. Syst. Technol. 2013, 21, 186–193. [CrossRef]
24. Palmieri, G.; Baric, M.; Glielmo, L.; Borrelli, F. Robust vehicle lateral stabilisation via set-based methods for uncertain piecewise
affine systems. Veh. Syst. Dyn. 2012, 50, 861–882. [CrossRef]
25. Fenyes, D.; Nemeth, B.; Gaspar, P. Analysis of autonomous vehicle dynamics based on the big data approach. In Proceedings of
the European Control Conference, Limassol, Cyprus, 12–15 June 2018; pp. 219–224.
26. Rajamani, R. Vehicle Dynamics and Control; Springer: Berlin/Heidelberg, Germany, 2005.
Energies 2021, 14, 517 16 of 16

27. Wang, Y.; Witten, I.H. Pace Regression; (Working Paper 99/12); University of Waikato, Department of Computer Science: Hamilton,
New Zealand, 1999.
28. Shibata, R. An optimal selection of regression variables. Biometrika 1981, 68, 45–54. [CrossRef]
29. Gill, P.E.; Murray, W.; Wright, M. Practical Optimization; Academic Press: London, UK, 1981.
30. Coleman, T.F.; Li, Y. A Reflective Newton Method for Minimizing a Quadratic Function Subject to Bounds on some of the
Variables. SIAM J. Optim. 1996, 6, 1040–1058. [CrossRef]