You are on page 1of 6

Preprints, 8th IFAC International Symposium on

Preprints,
Preprints, 8th
Advances IFAC
IFAC International
in Automotive
8th Control Symposium
International Symposium on
on
Preprints,
Advances 8th IFAC International Symposium on
June 19-23,in
Advances in Automotive
2016.
Automotive Control
Norrköping, Sweden
Control Available online at www.sciencedirect.com
Advances
June 19-23,in2016.
Automotive Control
Norrköping, Sweden
June
June 19-23,
19-23, 2016.
2016. Norrköping,
Norrköping, Sweden
Sweden
ScienceDirect
IFAC-PapersOnLine 49-11 (2016) 285–290
Vehicle
Vehicle Lateral
Lateral Motion
Motion Control
Control with
with
Vehicle
Vehicle Lateral
Lateral
Performance andMotion
Motion
SafetyControl
Control with
with
Guarantees
Performance
Performance
Performance and
and Safety
Safety
and Safety Guarantees
Guarantees
Guarantees
∗ ∗ ∗ ∗∗
L. Ni ∗ A. Gupta ∗ P. Falcone ∗ L. Johannesson ∗∗
L.
L. Ni ∗∗ A. Gupta ∗∗ P. Falcone ∗∗ L. Johannesson ∗∗ ∗∗
L. NiNi A. A. Gupta
Gupta P. P. Falcone
Falcone L. L. Johannesson
Johannesson

∗ Chalmers University of Technology - Department of Signals and
∗ Chalmers University of Technology - Department of Signals and
∗ Chalmers
Systems,
Chalmers University
Gothenburg,
University Swedenof Technology
Sweden
of Technology(e-mail:- Department of
of Signals and
Signals
leini@student.chalmers.se,
- Department and
Systems,
Systems, Gothenburg,
Gothenburg, Sweden
ankit.gupta@chalmers.se, (e-mail:
(e-mail: leini@student.chalmers.se,
leini@student.chalmers.se,
falcone@chalmers.se).
Systems,
∗∗
Gothenburg, Sweden
ankit.gupta@chalmers.se, (e-mail: leini@student.chalmers.se,
falcone@chalmers.se).
ankit.gupta@chalmers.se,
∗∗ Volvo Cars Corporation, Gothenburg,
ankit.gupta@chalmers.se, falcone@chalmers.se).
Sweden, (e-mail:
falcone@chalmers.se).
∗∗ Volvo Cars Corporation, Gothenburg, Sweden, (e-mail:
∗∗ Volvo lars.johannesson.mardh@volvocars.com)
Cars
Cars Corporation, Gothenburg, Sweden,
Corporation, Gothenburg,
Volvo lars.johannesson.mardh@volvocars.com) Sweden, (e-mail:
(e-mail:
lars.johannesson.mardh@volvocars.com)
lars.johannesson.mardh@volvocars.com)
Abstract: This paper explores the use of Model Predictive Control (MPC) techniques to solve
Abstract:
Abstract: This
This paper
paper explores the use of Model Predictive Control (MPC) techniques to solve
vehicle
Abstract:
vehicle
lateral
lateral paper explores
Thismotion
motion explores
control
the
the use
control problem
problemuse ofofon
on
Model
Model Predictive
highway scenarios.
Predictive
highway scenarios.
Control
ControlIn
(MPC)
(MPC) techniques
In particular,
particular,
the problem
techniques
the
to
to solve
problem solve of
of
vehicle
autonomously
vehicle lateral motion
driving
lateral driving a control
vehicle
motiona control problem
along
problem a on
desired highway
path
on highway isscenarios.
formulated,
scenarios. In particular,
where
Inwhere safety
particular, the problem
constraints
the problemand and of
of
autonomously
autonomously
performance driving
levels musta vehicle
vehicle
be along
along
guaranteed a
a desired
desired
for all path
path is
is
possible formulated,
formulated,
road where
curvatures safety
safety
within constraints
constraints
a compact and
set.
autonomously
performance driving
levels musta vehicle
be along
guaranteed a desired
for path is
all possible formulated, where safety constraints and
performance
Safety
performance levels
constraints
levelsare must
must be
be guaranteed
translated for
for all
into a maximum
guaranteed lateralroad
all possible
possible road curvatures
curvatures
deviation
road curvatures
within
within
and orientation a compact
within aa compacterror w.r.t.
compact
set.
set.
set.
aSafety
Safety
desired
Safety
constraints
constraints
path,
constraints
are
are
while
are
translated
translated
performance
translated
into
into
into
aa maximum
maximum
requirements
a maximum
lateral
lateral
are
lateral
deviation
deviation
formulated
deviation in and
and
terms
and
orientation
orientation
of bounded
orientation
error
error
error
w.r.t.
w.r.t.
lateral
w.r.t.
a
a desired
desired path,
and while performance requirements
simulation are formulated in terms of bounded lateral
desired path,
acceleration
a
acceleration path,
and
while
while performance
velocity. Preliminary
performance
velocity. Preliminary
requirements
requirements
simulation
are
results
are
results
formulated
show that
formulated
show
in
that
terms
in the
terms
the
of
of bounded
designed
designedbounded lateral
controller
lateral
controller
is
is
acceleration
capable
acceleration of and
delivering velocity.
and velocity. Preliminary
acceptable performance
Preliminary simulation
at
simulation theresults
cost
results ofshow
limited
show that the
online
thatonline designed
computational
the designed controller
costs.
controller is
is
capable
capable of
of delivering
delivering acceptable
acceptable performance
performance at
at the
the cost
cost of
of limited
limited online computational
computational costs.
costs.
capable of delivering acceptable performance at the cost of limited online
© 2016, IFAC (International Federation of Automatic Control) Hosting by Elsevier Ltd. All rights reserved. computational costs.
Keywords: Vehicle motion control, safety, model predictive control, autonomous driving
Keywords: Vehicle
Keywords: Vehicle motion control,control, safety, model model predictive control, control, autonomous driving driving
Keywords: Vehicle motion motion control, safety, safety, model predictive
predictive control, autonomous autonomous driving
1. INTRODUCTION proaches to the overall problem of vehicle motion control,
1. INTRODUCTION
1. INTRODUCTION proaches
proaches to
to the overall problem
from of vehicle motion
Measurecontrol,
1. INTRODUCTION climbing
proaches
climbing to the
the
the
ASIL
the
ASIL
overall
overall problem
ladder,
problem
ladder, from
of
of
a
vehicle
vehicle motion
a Quality
Quality motion control,
(QM)
control,
Among the technologies advancing within the automotive climbing climbing
to ASIL D, the will
the ASIL
ASIL leadladder,
to enormous,
ladder, from costly Measure
from aa Quality
Quality and time(QM)
Measure
Measure (QM)
con-
(QM)
Among the technologies
technologies advancing within the automotive
automotive to
to ASIL
ASIL
suming D,
D, will
will
verification lead
lead to
to
problems,enormous,
enormous,
which costly
costly and
and
inevitably time
time
will con-
con-
stall
Among
field,
Among the
autonomous
the driving
technologies advancing
is definitely
advancing within
within the
emerging
the with
automotive most to ASIL D, will lead to enormous, costly and time con-
field, autonomous driving is definitely emerging with most suming verification problems, which inevitably will stall
field,
promises
field, autonomous
of improving
autonomous driving
driving manyis definitely emerging
aspects emerging
is definitely with
of our lifestyles
with most re- suming verification problems, which inevitably will stall
most suming
the productverification problems,
development. In which
this inevitably
paper we focus will
on the
stall
promises of improving many aspects of our lifestyles re- the
the product
product development.
development. In
In this
this paper
paper we
we focus
focus on
on the
promises
lated
promises to of improving
transport.
of improving Road many
many aspects
safety, of
traffic
aspects our lifestyles
congestions
of our lifestylesand re-
and
re- theproblem
product
problem
of
of
designing
development.
designing
a
a
vehicle
In this
vehicle
lateral motion
papermotion
lateral we focus on the
controller
controllerthe
lated
lated
pollutant to
to transport.
transport.
emissions, Road Road safety,
transitsafety, traffic
traffic
efficiency, congestions
congestions
healthiness of and problem
with
ur- problem of designing
performancedesigning and aasafety
vehicle lateral motion
guarantees motion
and explorecontrollerthe
lated to transport. Road safety, traffic congestions and with of
performance and vehicle
safety lateral
guarantees and controller
explore the
pollutant emissions,totransittransit efficiency, healthiness of ur- with performance and
pollutant
ban emissions,
environments,
pollutant emissions,totransit name efficiency,
a few, are healthiness
efficiency, recognized of
healthiness be use
to ur-
of withof Model Predictive
performance and safetyControl
safety guarantees
guarantees and
and explore
(MPC) techniques, explorewith the
the
ban environments,
ban environments,
potentially and highlyto name
name
impacted a few,
a few,
by are
are recognized
recognized
autonomous to ur-
to
driving.be use
be use
the of
of Model
Model
objective Predictive
Predictive
of providing Control
Control
systematic (MPC)
(MPC)
design techniques,
techniques,
methodologies with
with
ban environments, to name a few, are recognized to be use of Model Predictive Control (MPC) techniques, with
potentially and highly
highly impacted by autonomous
autonomous driving. the the objective
objective of
of providing
D-type of systematic design methodologies
potentially
It is then natural
potentially and
and highly impacted
to question
impacted by
the maturity
by autonomous driving. to
driving.
of the available the
to
satisfy
objective
satisfy
ASIL
ASIL of providing
providing
D-type of
systematic
requirements.
systematic
requirements.
design
design methodologies
methodologies
It
It is
is then
then
autonomous natural
naturaldriving to
to question
question
technologies,the
the maturity
maturity
especially of
of the
the available
withavailable
respect to to satisfy
satisfy ASILASIL D-typeD-type of of requirements.
requirements.
It is then
autonomous naturaldriving to question
technologies,the maturity of
especiallyonwiththe
with available
respect In Guldner et al. (1996) steering control for passenger
autonomous
to the new and
autonomous driving
driving technologies,
demanding
technologies, especially
requirements
especiallyonwith the respect
vehicle In
respect In
cars Guldner
Guldner
on et
et
automated al.
al. (1996)
(1996)
highways steering
steering
is control
control
analyzed andfor
for passenger
passenger
to
to the
the
motion new
new
control and
and demanding
demanding
imposed requirements
by requirements
Level 4 NHTSA on the
the vehicle
vehicle
(2013) of In cars Guldner
on et al. (1996)
automated highways steering
is controland
analyzed for conditions
passenger
conditions
to the
motion new
control and demanding
imposed by requirements
Level 4 NHTSA on the vehicle
(2013) of cars
for the
cars on
on automated
safety
automated highways
and performance
highways is
is analyzed
criteria
analyzed areand
and conditions
proposed.
conditions In
motion
autonomous
motion control
control imposed
driving. by Level 4 NHTSA
imposed by Level 4 NHTSA (2013) of Lei (2013) of for
for the
the
et al. safety
safety
(2006) and
and performance
a performance
vision-based lane criteria
criteria are
are
detection proposed.
method In
proposed. In
is
autonomous
autonomous driving.
driving. for
Lei the
et safety
al. (2006) and performance
aa avision-based criteria
lane are
detection proposed.
method In
is
autonomous
While the existingdriving. vehicle motion controllers use the driver utilized Lei
Lei et al.
et al. (2006)
along
(2006)with vision-based
PID controller
a avision-based lane detection
for the
lanefordetectionlateral method
control.
method is
is
While the existing vehicle motion controllers use the driver utilized
utilized
A along
along
comparative with
with
studya PID
PID
of controller
controller
linear for
controllers the
the lateral
lateral
for lane control.
control.
keeping
While
as a
While the
the existing
failsafe
existing vehicle
fall-back,
vehiclein motion
autonomous
motion controllers
driving,
controllers use
use the
the
the driver
devia-
driver utilized along with a PID controller for the lateral control.
as aa form
failsafe fall-back, inforautonomous driving, the devia-
devia- A A comparative
cancomparative
be found instudy study
Taylor of et
of linear
linear controllers
controllers
al. (1999). for lane
for
A dynamic lane keeping
keeping
as
tion
as failsafe
a form
failsafe fall-back,
a given path,in
fall-back, inforautonomous
example, must
autonomous driving,
driving, the
be guaranteed
the devia- A cancomparative
be found in study
Taylor of et
linear
al. controllers
(1999). A for lanefeedback
dynamic keeping
feedback
tion
tion
to form
satisfy a given
aaangiven path,
path,
Automotive for example,
example,
Safety must
must
Integrity be
be guaranteed
guaranteed
Level can be
controller foundis in Taylor
proposed inet al. (1999).
Benine-Neto
(ASIL) can be found in Taylor et al. (1999). A dynamic feedback A et dynamic
al. (2010),feedback
which
tion form
to satisfy
satisfy an given
an see path,
Automotive for example,
Safety must
Integrity be guaranteed
Level (ASIL)
(ASIL) controller is proposed in Benine-Neto et al. (2010), which
to
requirement, Automotive
ISO-26262 Safety
(2011) Integrity
for a Level
description of the controller
considers
controller is
roadproposed
curvature
is proposed in Benine-Neto
as bounded
in Benine-Neto et al. (2010),
disturbance
etdisturbance which
input.
al. (2010), input.
which
to satisfy an see
requirement, Automotive
ISO-26262 Safety
(2011) Integrity
for a Level (ASIL)
description of the considers road curvature as bounded
requirement,
ASIL standard. see ISO-26262
The (2011)
determination for
of a description
the required of the
ASIL considers
Since
considers the road
vehicle
road curvature
motion
curvature as
as bounded
dynamics
bounded disturbance
are nonlinear,
disturbance input.
con-
input.
requirement,
ASIL standard. see ISO-26262
The determination (2011)
determination for a
of thedescription
the required Smithof the Since the
ASIL straints the vehicle motion dynamics are can nonlinear, con-
ASIL
is the standard.
result of The
hazard analysis and of
risk required
assessment ASIL Since
Since the vehicle
related
vehicleto motion
safety
motion and dynamics
performance
dynamics are
are nonlinear,
be
nonlinear, con-
naturally
con-
ASIL
is the standard.
result of The determination
hazard analysis and of the
risk required Smith
assessment ASIL straints related to safety and performance can be naturally
is
and the result
Simpson of hazard
(2010), analysis
which means and risk assessment
that Smith
functionalities straints
accommodated
straints related
related to
with
to safety
MPC
safety and
and performance
techniques,
performance like can
in
can be
be naturally
Falcone et
naturally al.
is the result
and Simpson
Simpson of hazard analysis
(2010),forwhich
which and
means risk assessment
that functionalities Smith
functionalities accommodated with MPC techniques, like in Falcone et al.
and
with likely (2010),
potential means
severely that
life-threatening or fatal accommodated
(2007),
accommodated where a with
MPC
with MPC
MPC techniques,
strategy for
techniques, like
steering
like inin Falcone
control
Falcone of et
et al.
vehi-
al.
and
with Simpson
likely (2010),forwhich
potential means
severely that functionalities
life-threatening or fatal (2007), where a MPC strategy for steering control of vehi-
with
injury likely
in the potential
event offor
a severely
malfunction life-threatening
will be or
classified fatal
as (2007),
cle on
(2007), where
slippery
where aroada MPC
road
MPC is strategy
proposed.
strategy forInsteering
Lee
forInsteeringet control
al. (2012)
control of avehi-
fast
ofavehi-
with likely
injury D, potential
in requiring
the eventevent of for severely
of aavehicle
malfunction life-threatening
will be or
be classified
classified fatal
as MPCcle on slippery is proposed. Lee et al. (2012) fast
injury
ASIL in the malfunction will as cle on slippery
slippery road
on strategy is
is proposed.
is proposed In
In Lee
for lateral et
et al.
al. (2012)
Leecontrol. a
a fast
The paper
injury
ASIL D, the eventthe
in requiring of
the avehicle manufacturer
malfunction will be to
manufacturer to
guarantee
classified
guarantee as cleMPC strategy road
is proposedproposed.
for lateral control. (2012)
The fast
paper
ASIL
aASIL D,
failure requiring
rate of 10 the−8vehicle manufacturer
eventsmanufacturer to guarantee
per hour. Intoconclusion, MPC
proposes
MPC strategy
an
strategy is
algorithm
is proposed
proposedto for
for lateral
approximate
lateral control.
solution
control. The
of the
The paper
opti-
paper
D, requiring the−8vehicle guarantee proposes an algorithm to approximate
aa failure
a failure
in rate
Level 4rate
failure rate
of 10
of
autonomous 10−8 events
events
events per
of 10−8 driving,
per hour.
hour.
staying
per
In
In conclusion,
In conclusion,
hour.within conclusion,
proposes
lane will mization
proposes an
an algorithm
problem
algorithm to the MPCsolution
to approximate
underlying approximate solution
controller,
solution
of
of the
of the
opti-
opti-
by using
the opti-
in Level 4 autonomous driving, staying within lane will mization problem
mization problem
precomputed problem underlying
underlying
solutions. A MPCthe MPC
the MPC
MPC controller,
controller,
controller is by using
by using
designed using
to
in
need
in Level
to
Level 4
be
4 autonomous
guaranteed
autonomous driving,
with
driving,ASILstaying
D.
staying within
Hence,
within lane
the
lane will
ASIL
will mization underlying the controller, by
need to be guaranteed with ASIL D. Hence, the ASIL precomputed
precomputed
resemble the solutions.
solutions.
driver A
A
behavior MPC
MPC in controller
controller
Gray et is
is
al. designed
designed
(2012). Theto
to
need
D
need to
to be
requirement
be guaranteed
to
guaranteed stay with
within
with ASIL
the
ASIL D.
lane
D. Hence,
will
Hence, the
reflect
the ASIL
into
ASIL precomputed solutions. A MPC controller is designed to
D requirement to stay within the lane will reflect into resemble
resemble
controller the
the
is driver
driver
designed behavior
behavior
to only in
in
apply Gray
Gray
the et
et al.
al.
correcting(2012).
(2012). The
The
control
D
D requirement
a stringent
requirement to
requirement
to stay
stay within
within the
on the maximum
the lane
lane will
will reflect
totalreflect
deviationinto
into resemble the driver behavior in Gray et al. (2012). The
a stringent requirement on the maximum total deviation controller
controller thatisis designed
necessaryto
designed to only
only apply
apply the
the correcting control
a
from
a stringent
the desired
stringent requirement
requirement on
on the
trajectory/path the maximum
(performance
maximum total
totaland safety action
deviation
deviation controller
action that isis
isdesigned
necessary toto
to
avoid
only
avoidapply the correcting
violation
violation
of the safety
correcting
of the
control
con-
control
safety con-
from
from the
the
guarantees), desired
desired trajectory/path
trajectory/path
compatibly with the (performance
(performance
sensing and
and
technology. safety
safety
It is action that
straints. that
A is necessary
MPCnecessary
problem to avoid
avoid
for violation
obstacle of the
avoidancethe safety
safety
and con-
lane
from the desired trajectory/path (performance and safety action
straints. A is
MPC problem to for violation
obstacle of
avoidance and con-
lane
guarantees),
guarantees),
then clear that, compatibly
compatibly
without with
with the
the
systematic sensing
sensing
control technology.
technology.
engineering It
It is
is
ap- straints.
keeping A
is MPC
proposed problem
in Turrifor obstacle
et al. avoidance
(2013), based and
on lane
linear
guarantees), compatibly with the sensing technology. It is straints.
keeping A
is MPC
proposed problem
in Turrifor obstacle
et al. avoidance
(2013), based and
on lane
linear
then
then clear
clear that,
that, without
without systematic
systematic control
control engineering
engineering ap-
ap- keeping
decoupled is proposed
lateral and in Turri et
longitudinal al. (2013),
dynamics, basedthus on linear
helping
then clear that, without systematic control
 This work is partially supported by the Vinnova FFI Complex engineering ap- keeping
decoupled is proposed
lateral and in Turri et al. (2013),
longitudinal dynamics, based on linear
thus helping

decoupled
decoupled lateral lateral andand longitudinal
longitudinal dynamics,
dynamics, thus thus helping
helping
 This work is partially supported by the Vinnova FFI Complex
Control
 This Program, under the grant No. 2015-02309.
This work
Control work is
is partially
Program,partially
under
supported
supported
the grant
by the
the Vinnova
Vinnova FFI
by2015-02309.
No. FFI Complex
Complex
Control Program, under the grant No. 2015-02309.
Control Program, under the grant No. 2015-02309.
Copyright
2405-8963 ©© 2016,
2016 IFAC
IFAC (International Federation of Automatic Control)292 Hosting by Elsevier Ltd. All rights reserved.
Copyright
Copyright
Peer review©
© 2016
2016
under IFAC
IFAC
responsibility of International Federation of 292
292
Automatic
Copyright © 2016 IFAC 292Control.
10.1016/j.ifacol.2016.08.043
IFAC AAC 2016
286
June 19-23, 2016. Norrköping, Sweden L. Ni et al. / IFAC-PapersOnLine 49-11 (2016) 285–290

in framing a convex QP problem for fast calculation of the values, can be approximated as,
solution. vy + lf ψ̇ vy − lr ψ̇
αf = − δ, αr = , (3)
In this paper, we explore MPC approaches to the vehicle vx vx
lateral motion control problem. We focus on the problem where δ denotes the front steering angle as depicted in
of controlling the lateral vehicle motion subject to safety Figure 1. In order to use the steering rate as control input,
and performance requirements, along low curvature paths
like in, e.g., highways. The road curvature is considered as
disturbance input to the system. Safety and performance
requirements are formulated in terms of the maximum
deviation from the desired path and constraints on the
vehicle states stemming from a desired comfort envelope.
These constraints are guaranteed to be persistently satis-
fied within a known set of vehicle states for a curvature of
the desired path within given boundaries. Preliminary sim-
ulation results show the performance and the viability of
the proposed approach, encouraging further developments.
The paper is structured as follows. In Section 2 we intro-
duce a vehicle model, notations, and formally state the
vehicle lateral motion control problem. Section 3 presents
few preliminary results on invariant set and an algorithm
to calculate the invariant set. In Section 4, the design Fig. 1. Vehicle in a desired path based coordinate system
procedure is shown. while Section 5 show the results of
the model (1) is augmented with an integrator. Hence, for
numerical simulations. The paper is concluded is in Sec-
a given vehicle longitudinal speed vx , the model (1)-(3)
tion 6 with final remarks about the presented results and
can be compactly written as,
future research directions.
ẋ(t) = Ax(t) + Bu(t) + Ew(t), (4)
2. PROBLEM FORMULATION  T
where x = vy , ψ̇, eψ , ey , δ and w = γ are the state
2.1 Vehicle Modeling and the disturbance vectors and u = δ̇ is the steering input
command.
Consider the vehicle model sketched in Figure 1. For small
road bank angle, the vehicle motion w.r.t. the path Γdes , 2.2 System Constraints
subject to the lateral and yaw dynamics, is described
by the following set of differential equations (Rajamani, The input, state and disturbance vector in (4) is subject to
2006). a set of physical and design constraints. These constrains
  are the result of safety, performance and physical limita-
mv̇y = −mvx ψ̇ + 2 Fyf + Fyr , (1a)
tion of a vehicle. The safety requirements, for the consid-
Jz ψ̈ = 2[lf Fyf − lr Fyr ], (1b) ered problem, translate into the following constraints on
ėψ = ψ̇des − ψ̇, (1c) the position ey
ėy = −vy + vx eψ , (1d) eymin ≤ ey ≤ eymax , (5)
ψ̇des = vx γ, x (1e) To preserve the driving comfort, we impose bounds on the
where m and Jz denote the vehicle mass and yaw inertia, lateral vehicle speed and acceleration, which, for a given
respectively, lf and lr are the distances of the vehicle center speed vx , can be written as,
of gravity from the front and rear axles, respectively, vx vymin ≤ vy ≤ vymax , (6a)
and vy are the longitudinal and lateral velocities, respec- aymin aymax
tively, in the vehicle body frame, ψ̇ is the turning rate, ≤ ψ̇ ≤ , (6b)
vx vx
where ψ denotes the vehicle orientation w.r.t. the fixed Further physical constraints stem from the limited steering
global frame (X, Y ) in Figure 1. Fyf , Fyr are the lateral and steering rate of the steering actuator.
tire forces at the front and rear axles, respectively. In (1c) δmin ≤ δ ≤ δmax ,
and (1d), eψ and ey denote the vehicle orientation and (7)
position, respectively, w.r.t. the path Γdes and ψdes is the δ̇min ≤ δ̇ ≤ δ̇max .
desired vehicle orientation, i.e., the slope of the tangent to The constraints (5)-(7) can be compactly rewritten for the
the path Γdes in the point O. system (4) as,
The lateral tire forces in (1a) and (1b) are generated at the X = {x ∈ R4 : Hx x ≤ hx },
(8)
tire contact patch and are, in general, nonlinear functions U = {u ∈ R : Hu u ≤ hu }.
of the vehicle states. In this paper, we compute the lateral
tire forces as, Finally, we assume that the curvature γ of the reference
Fyi = −Ci αi , i ∈ {f, r}, (2) path Γdes is bounded, i.e., it belongs to the set,
where Ci are the tire cornering stiffness coefficients at the
two axles and αi are the tyre slip angles which, for small W = [γmin , γmax ] . (9)

293
IFAC AAC 2016
June 19-23, 2016. Norrköping, Sweden L. Ni et al. / IFAC-PapersOnLine 49-11 (2016) 285–290 287

2.3 Problem Statement definitions are derived from Blanchini (1999); Bertsekas
and Rhodes (1971); Bertsekas (1971); Kolmanovsky and
Consider the problem of regulating the states of the system Gilbert (1998).
for a given constant speed vx . Definition 2. (Robust Control Invariant Set). A set C ⊆
x(t + 1) = Ad x(t) + Bd u(t) + Ed w(t), (10) X is a robust control invariant set for the system in (13)
subject to the constraints (8), (9). Where Ad , Bd and Ed subject to the constraints in (14), if
are the discrete versions of system matrices in (4). The x(t) ∈ C ⇒ ∃ u(t) ∈ U such that f (x(t), u(t), w(t)) ∈ C,
problem can be formulated as a finite time, constrained ∀w(t) ∈ W, ∀t ∈ N+
optimal control problem, solved in receding horizon. In
(16)
particular, the closed-loop, state feedback control law.
ucl (x(t)) = u∗t|t (x(t)) , (11) A robust control invariant set can be calculated by using
is adopted, where u∗t|t
(x(t)) is found as the solution of the the condition provided by the following theorem Kerrigan
(2000); Dórea and Hennet (1999)
following optimization problem.
Problem 1 Theorem 1. (Condition for invariance). A set C ⊆ X is a
−1
t+N robust control invariant set for the system (13) subject to
Jt∗ (x(t)) = min xN |t 2P + xk|t 2Q + uk|t 2R constraints (14), if and only if
Ut
k=t Pref (C, W) ∩ C = C (17)
(12a) Definition 3. (Maximal Robust Control Invariant Set C∞ ).
subj. to The set C∞ is the maximal robust control invariant set for
xk+1|t = Ad xk|t + Bd uk|t + Ed wk|t , (12b) the system in (13) subject to the constraints in (14), if it
xk|t ∈ X , uk|t ∈ U, wk|t ∈ W, (12c) is robust control invariant and contains all robust control
invariant sets contained in X .
xt+N |t ∈ Xf , (12d)
xt|t = x(t), (12e) 3.2 Calculation of RCI
 
where Ut = ut|t , . . . , ut+N −1|t , N is a finite time horizon,
P, Q  0 and R  0 are weighting matrices of appropriate In order to calculate the robuust control invariant set for
sizes. Further, Xf is a terminal constraint, selected as a Ro- system (12b) we have proposed an algorithm. The set C
bust Control Invariant Set, such that persistent feasibility in (17) can be used, together with the definition of the
for Problem 1 can be guaranteed (Kerrigan, 2000). The set Pref in equation (18), written for the model in (12b)
construction of this set is explained in Section 4. In order and the sets (8) and (9). In particular, the set C can be
to assist the construction of terminal constraints we need calculated by the following algorithm
to recall few definition on invariant sets in Section 3. Also,
an approach is proposed for the construction of robust Algorithm 1 Computation of the robust control invariant
control invariant (RCI) set. set C
INPUT: X , W, Ad , Bd , Ed
3. INVARIANT SET OUTPUT: C
1: C0 ← X
3.1 Preliminaries 2: k ← 0
3: Ck+1 ← Pref (Ck , W) ∩ Ck
In this section we introduce a few definitions and recall 4: if Ck+1 = Ck , then
basic results on set invariance theory and reachability 5: goto 10,
analysis for constrained systems and provide the results 6: else
on feasibility of MPC schemes used next in this paper. A 7: C k ← Ck+1
comprehensive survey of papers on set invariance theory 8: k ← k + 1,
can be found in Blanchini (1999). 9: goto 3.
10: return Ck
We will denote the set of all real numbers and positive
integers by R and N+ , respectively.
where, the set Pref at step 3 is calculated as
For the system,
Pref (X , W)  {x ∈ X |
x(t + 1) = f (x(t), u(t), w(t)), (13)
∃ u ∈ U s.t. Hx Ad x + Hx Bd u (18)
subject to the constraints
≤ min hx − Hx Ed w}.
x(t) ∈ X , u(t) ∈ U ⊆ Rm , w(t) ∈ W. (14) w∈W

Definition 1. (Pref set). we define the set of states which Since the set Ck+1 is the result of a sets intersection,
can be driven into the target set S in one time step as the complexity of its representation, i.e., the number of
Pref (S, W)  {x ∈ Rn | inequalities, increases at every iteration. Since Algorithm 1
∃ u ∈ U s.t. f (x, u, w) ∈ S, ∀w ∈ W}. stops as the condition at step 4 is verified, the represen-
(15) tation of the set C might consist of an unnecessarily high
number of inequalities, thus complicating the Problem 1.
The Pref set introduced in Definiton 1 can be used to In this paper, Algorithm 1 is stopped when the Euclidean
calculate the robust control invariant sets. The following distance between the Chebyshev’s centers of Ck+1 and Ck ,

294
IFAC AAC 2016
288
June 19-23, 2016. Norrköping, Sweden L. Ni et al. / IFAC-PapersOnLine 49-11 (2016) 285–290

respectively, is below a given threshold. That is, the condi-


tion at step 4 is replaced with the following condition rk −
rk+1  ≤ rthr , where rk denotes the Chebyshev’s center
of Ck and rthr is a threshold. Further stopping conditions
can be added like, e.g., monitoring the radius of the Chebi-
shev’s ball enclosed in Ck+1 and Ck .We show next how
robust control invariant sets C can be used to guarantee
feasibility of MPC schemes.

4. CONTROL DESIGN

Since a MPC controller iteratively solves optimization


Problem 1, persistent feasibility has to be guaranteed.
Persistent feasibility implies that, if Problem 1 is feasible
(a) Volume.
for some initial x(0) ∈ X0 , where X0 is the set of initial
states for which the Problem 1 is feasible, the state
trajectory is guaranteed to evolve within X0 . Persistent
feasibility cannot be guaranteed for any choice of the
tuning parameters. As mentioned in Kerrigan (2000), by
selecting Xf as a robust control invariant set persistent
feasibility can be guaranteed i.e.
xt+N |t ∈ C, (19)
where C is a robust control invariant set mentioned in
Definition 2. Results enforcing robust closed-loop stability
are available for min-max schemes Kerrigan (2000). In
this paper, we enforce closed-loop stability condition for
the nominal system only, i.e., with w = 0 in (12b).
Although our approach does not guarantee stability for all
possible road curvatures, extensive simulations have shown
(b) 2-norm of the difference between consecutive radii
stable operation of the closed loop system for severe road of the Chebyshev ball.
curvature changes.
The terminal cost term in the cost function (12a), Fig. 2. Calculation of the maximal robust control invariant set.
−1
t+N
rameters used in the vehicle model (1) are reported in
Jt (x(t), Ut ) = xt+N |t 2P + xk|t 2Q +uk|t 2R , (20) Table 1. The following bounds have been used in the
k=t
where the matrix P is the solution of the following Alge- Table 1. Model parameters
braic Riccati Equation (ARE)
  −1 T  Parameter Description Value
P = ATd P − P Bd BdT P Bd + R Bd P Ad + Q. (21) m Mass 2164 [kg]
Jz Yaw moment of inertia 4373 [kg × m2 ]
Furthermore, instead of (19) we add the following con- Cr Rear cornering stiffness coeff. 122380 [N/rad]
straint, Cf Front cornering stiffness coeff. 150540 [N/rad]
xt+N |t ∈ O, (22) lr Rear axle to CoG distance 1.6456 [m]
lf Front axle to CoG distance 1.3384 [m]
where O is the maximal robust invariant set for the closed-
loop system
constraints (5)-(7) to construct the sets X , U in (12)
x(k + 1) = (Ad + Bd K) x(k) + Ed w(k), (23) 
and K is the Linear Quadratic (LQ) regulator. The robust 
 ey = −eymin = 0.2 [m],
 max

invariant set O can be calculated with an algorithm similar 
 v ymax = −vymin = 0.4 [m/s],

 ◦
to Algorithm 1, after re-defining the set Pref at step 3 as 
eψmax = −eψmin = 5[ ] = 0.0873 [rad],
aymax = −aymin = 3[m/s2 ], (25)


Pref (X , W) {x ∈ X |Hx (Ad + Bd K) x  ψ̇
 max = − ψ̇min = 8.251[ ◦
]/s = 0.144 [rad/s],
(24) 

≤ min hx − Hx Ed w}.  ◦
δmax = −δmin = 2.5[ ] = 0.0436 [rad],
w∈W 

δ̇max = −δ̇min = 2.86 [◦ ]/s = 0.05 [rad/s].
Remark 1. Here we only focus on nominal stability i.e for
w = 0. Robust stability can be acheived by introducing Furthermore the set (9) is defined by the following bounds
changes in road curvature as additional variable ∆w = on the road curvature.
w(k + 1) − w(k) and using the same framework. γmax = −γmin = 0.0263.
The constraints in (25) may look more restrictive than nec-
5. SIMULATIONS essary for normal driving conditions. Nevertheless, we re-
call that we consider high-speed, highway scenarios where
The controller (12) has been tested in simulations and an autonomous driving system should deliver a comfort-
compared against a LQR regulator. The physical pa- able driving experience. The matrices Ad , Bd , Ed in (12b)

295
IFAC AAC 2016
June 19-23, 2016. Norrköping, Sweden L. Ni et al. / IFAC-PapersOnLine 49-11 (2016) 285–290 289

have been obtained from the model (4) for vx = 75 [km/h],


by applying the Zero-Order Hold (ZOH) discretization
method, with a sampling time Ts = 25 [ms]. The weighting
matrices of the cost (12a) have been chosen as follows
 
40 0 0 0 0
0 1 0 0 0
Q =  0 0 1 0 0  , R = 2.
 
0 0 0 1 0
0 0 0 0 20
With this choice of the tuning parameters, the LQ gain
in (23) and the corresponding solution of the ARE (21)
are, respectively,
(a) Lateral deviation.
K = [3.6538 − 0.5135 27.9792 − 1.5585 − 14.6065],
 
0.0588 −0.0057 0.2098 −0.0040 −0.0360
−0.0057 0.0011 −0.0273 0.0001 0.0050 
P = 104 ×  0.2098 −0.0273 1.1977 −0.0358 −0.2660 .
 
−0.0040 0.0001 −0.0358 0.0036 0.0145
−0.0360 0.0050 −0.2660 0.0145 0.1311
In particular, the matrix P has been used as terminal
cost in (20). Finally, the terminal constraint (22) is en-
forced, where the invariant set O is calculated as in Algo-
rithm 1 with the set Pref defined as in (24). Figure 2
shows the evolution of the Euclidean distance between
the Chebyshev’s centers of Ok+1 and Ok and the radius
of the Chebyshev’s ball enclosed in Ok . In this work,
Algorithm 1 is stopped by using the stopping criterion (b) Orientation error
mentioned in Section 3.2 and by monitoring the volume
of Ok . In Figure 3 the curvature of the desired path Γdes

(c) Yaw rate

Fig. 4. Safety constraints. Deviation of the vehicle from the desired


Fig. 3. Curvature of the desired path. path with the LQ and the MPC controller.
a LQ controller shows improvements in the constraints
used in our simulation is reported. A comparison of the satisfaction at the cost of reasonable on-line computational
LQR and the MPC controllers is shown in Figures 4-6, complexity.
where the control input generated by LQR controller has
been clipped according to the control input constraints Although preliminary, this work is the basis for further
in (25), before applying it to system. The simulations show investigations aiming at understanding the impact on
that the clipped LQR controller is unable to stabilize the the conservativeness and the complexity of the controller
system, thus leading to unstable oscillations. The MPC of a design guaranteeing the satisfaction of safety and
controller, on the other hand, allow to easily accommodate performance requirements, despite of model uncertainty
closed- loop stability and design and physical constraints. and measurement noise.

6. CONCLUSIONS REFERENCES
Benine-Neto, A., Scalzi, S., Mammar, S., and Netto, M.
This paper presents preliminary simulation results of a (2010). Dynamic controller for lane keeping and obstacle
MPC-based the lateral vehicle motion controller. In par- avoidance assistance system. IEEE International Con-
ticular, the controller aims at minimizing the lateral de- ference on Intelligent Transportation Systems, 1363–
viation and the orientation error w.r.t. a desired path, 1368.
while guaranteeing safety and performance requirements. Bertsekas, D.P. (1971). Control of Uncertain Systems with
Namely, bounding the deviation form the path and the a set–membership description of the uncertainty. Ph.D.
lateral acceleration, respectively. A comparison against thesis, Electronic Systems Laboratory, MIT.

296
IFAC AAC 2016
290
June 19-23, 2016. Norrköping, Sweden L. Ni et al. / IFAC-PapersOnLine 49-11 (2016) 285–290

Blanchini, F. (1999). Set invariance in control — a survey.


Automatica, 35(11), 1747–1768.
Dórea, C. and Hennet, J.C. (1999). (a, b)-invariant
polyhedral sets of linear discrete-time systems. Journal
of Optimization Theory and Applications, 103(3), 521–
542.
Falcone, P., Borrelli, F., Asgari, J., Tseng, H.E., and
Hrovat, D. (2007). Predictive active steering control
for autonomous vehicle systems. IEEE Transactions on
Control Systems Technology, 15, 566–580.
Gray, A., Ali, M., Gao, Y., Hedrick, J.K., and Borrelli,
F. (2012). Integrated threat assessment and control
(a) Lateral velocity. design for roadway departure avoidance. IEEE Inter-
national Conference on Intelligent Transportation Sys-
tems, 1714–1719.
Guldner, J., Tan, H., and Patwardhan, S. (1996). Analysis
of automatic steering control for highway vehicles with
look-down lateral reference systems. Vehicle System
Dynamics, 26(4), 243–269.
ISO-26262 (2011). Road Vehicles Functional Safety, ISO
26262–Part–9, Geneva, Switzerland.
Kerrigan, E.C. (2000). Robust Constraints Satisfaction:
Invariant Sets and Predictive Control. Ph.D. thesis,
Department of Engineering, University of Cambridge,
Cambridge, England.
Kolmanovsky, I. and Gilbert, E.G. (1998). Theory and
(b) Lateral acceleration. computation of disturbance invariant sets for discrete-
time linear systems. Mathematical Problems in Egineer-
Fig. 5. Performance (comfort) constraints satisfaction of the LQ ing, 4, 317–367.
and the MPC controller. Lee, S.H., Lee, Y.O., Kim, B.A., and Chung, C.C. (2012).
Proximate model predictive control strategy for au-
tonomous vehicle lateral control. American Control
Conference, 3605–3610.
Lei, G., Jianqiang, W., and Keqiang, L. (2006). Lane keep-
ing system based on thasv-ii platform. IEEE Interna-
tional Conference on Vehicular Electronics and Safety,
305–308.
NHTSA (2013). U.S. Department of Transportation Policy
Release on Automated Vehicle Development.
Rajamani, R. (2006). Vehicle Dynamics and Control.
Springer.
Smith, D. and Simpson, K. (2010). Safety Critical Sys-
tems Handbook - A Straightforward Guide to Functional
(a) Steering rate. Safety, IEC 61508 and Related Standards.
Taylor, C., Kosecka, J., Blasi, R., and Malik, J. (1999). A
comparative study of vision-based lateral control strate-
gies for autonomous highway driving. The International
Journal of Robotics Research, 18(5), 442–453.
Turri, V., Carvalho, A., Tseng, H.E., Johansson, K.H.,
and Borrelli, F. (2013). Dynamic controller for lane
keeping and obstacle avoidance assistance system. IEEE
International Conference on Intelligent Transportation
Systems, 378–383.

(b) Steering angle.

Fig. 6. Control signal of the LQ and the MPC controller.

Bertsekas, D.P. and Rhodes, I.B. (1971). On the minimax


reachability of target sets and target tubes. Automatica,
7, 233–247.

297

You might also like