2014-01-9128

Published 08/01/2014
Copyright © 2014 SAE International
doi:10.4271/2014-01-9128
saepcmech.saejournals.org

Design and Testing of ABS for Electric Vehicles with Individually

Controlled On-Board Motor Drives
Valentin Ivanov, Barys Shyrokau, and Dzmitry Savitski
Ilmenau University of Technology

l
Javier Orus, Rubén Meneses, and José-Manuel Rodríguez-Fortún

na
Instituto Tecnológico de Aragón

Johan Theunissen and Karel Janssen

Flanders' DRIVE

io
ABSTRACT
The paper introduces the results of the development of anti-lock brake system (ABS) for full electric vehicle with

at
individually controlled near-wheel motors. The braking functions in the target vehicle are realized with electro-hydraulic
decoupled friction brake system and electric motors operating in a braking mode.

The proposed ABS controller is based on the direct slip and velocity control and includes several main blocks for

rn
computing of predictive (feedforward) and reactive (feedback) brake torque, wheel slip observer, slip target adaptation,
and the algorithm of brake blending between friction brakes and electric motors.

The functionality of developed ABS has been investigated on the HIL test rig for straight-line braking manoeuvres on
te
different surfaces with variation of initial velocity. The obtained experimental results have been compared with the
operation of baseline algorithm of a hydraulic ABS and have demonstrated a marked effect in braking performance.

CITATION: Ivanov, V., Shyrokau, B., Savitski, D., Orus, J. et al., "Design and Testing of ABS for Electric Vehicles with
In

Individually Controlled On-Board Motor Drives," SAE Int. J. Passeng. Cars - Mech. Syst. 7(2):2014, doi:10.4271/2014-01-9128.
E

INTRODUCTION Nm). The tests have confirmed an improved ABS performance

with a higher actuation frequency in the case of blended
Full electric vehicles with individual drive motors for each wheel
actuation of the friction brakes and electric motors as
SA

belong to the most promising technologies of next-day mobility.

compared with commonly-used hydraulic ABS, where
Such powertrain architecture not only provides energy-efficient
modulation is usually around 9 Hz. In particular, it was
driving but also opens new embodiment variants of traditional
mentioned that ABS function realized through electric motor
chassis systems contributing to enhancement of active safety
actuation can reduce the braking distance up to 7 % on ice
and driver comfort. This is particularly true for ABS, which
road as compared with a conventional hydraulic ABS.
becomes an opportunity of wheel torque modulation both
through friction brakes and electric motors. In that case the
Another industrial implementation of ABS with electric motors
ABS operation executes active safety functions and
©

on a full-scale car is given in [3]. This variant concerns the

simultaneously improves the electric driving efficiency due to
roadster electric vehicle (mass 1250 kg) equipped with
regenerative braking.
in-wheel electric motors of a maximum continuous torque 500
Nm and continuous power 73 kW. In this configuration, rear
An analysis of state-of-the-art in ABS with electric motor
electric motors in braking mode can guarantee minimum
actuation points to several practical solutions is worth
vehicle deceleration 0.3 g on the road surfaces with the friction
mentioning. For instance, the works [1, 2] have introduced
coefficient ≥ 0.3. In the case of higher deceleration demands,
results of the ABS operation realized through individual
the front friction brakes are activated.
in-wheel motors on full electric vehicle of executive class
(vehicle mass 2100 kg; maximum vehicle velocity 200 km/h;
maximum wheel power 40 kW; maximum wheel torque 550

902
 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014) 903

Both mentioned vehicle configurations relate to the powertrain simulation for road and operational conditions. Full-scale
architecture with in-wheel motors. It should be pointed out that experiments on real vehicles are rarely presented. Most of
in-wheel motors have still high costs despite continuous founded research publications consider the vehicles with
improvement of relevant design and production technologies. in-wheel motors, and there is a lack of deep studies analyzing
However, modern techniques propose several other variants of advantages, performance and functional limitations of anti-lock
the architecture. Of particular interest is electric vehicle braking systems with regard to electric powertrain architecture
configuration with near-wheel (or on-board) electric motors consisting of near-wheel motors, gearboxes and half-shafts.
connected to wheels through gearbox and half-shaft, Figure 1. In
such case the torsional dynamics of the half-shaft can bring Taking into account that the presented study is aimed at the

l
potential limitation in the actuation of ABS that was shown and implementation of the ABS on a real sport utility vehicle and
discussed in [4, 5]. To investigate this problem in more details,

na
should be subjected to experimental validation, it was decided
the main goal of the presented study has been formulated as the to propose robust ABS realization using proportional-integral
development and validation of the control strategy for anti-lock direct slip control with feedforward and feedback slip control
brake system to be implemented on an electric vehicle with parts. Next sections of the paper introduce the logic of the
individually controlled on-board motors connected to wheels discussed ABS in the application to specific electric vehicle
through powertrain elements with distinct torsional dynamics.

io
architecture and particular results of real-time validation of the
ABS algorithm on the hardware-in-the-loop (HIL) test rig.

at
ABS CONTROLLER
General Structure
The proposed architecture of anti-lock brake system can be

Figure 1. The powertrain architecture with on-board motor

The choice of a basic control method for the ABS has been
rn explained with Fig. 2 (as applied to single wheel control).

The driver control action Fdriver is the input of base brake

controller generating the torque demand Tdem for each wheel,
te
done from analysis of different solutions presented in recent which is added up with the reactive torque Treact. The reactive
research publications. The majority of the relevant ABS torque is computed by the reactive torque controller, which
investigations consider a simultaneous operation of both uses the reference slip ratio λref, the observed slip ratio λ and
the vehicle velocity Vx as input parameters. The reference slip
In

friction and motor variants due to strict vehicle safety

requirements. The ABS architecture thus has a blended ratio is produced by the reference slip generator, and the
composition in which the continuous of friction braking and parameters λ and Vx are computed with slip and velocity
negative motor torques depend on operational conditions and observers correspondingly. The sum of torques Tdem and Treact
the braking dynamics. For example, the work [6] introduces a is further processed in the block defining torque blending. This
variant of the blending using a rule-based controller. The block splits the total wheel torque demand into the torque
E

actuation of electric motors and hydraulic brake system in this demand for the electric motors Tem_dem and for the friction
ABS is governed by a set of rules that operate with slip brakes Tbr_dem. Both torques are being realized by
thresholds and four levels of specific vehicle deceleration. The corresponding actuators and the actuator outputs Tem and Tbr
SA

study [7] has proposed to use Sliding Mode Control for the are further used either in the vehicle model (in the case of
blended ABS operation. Another approach is presented in [8] in software- and hardware-in-the-loop simulations) or the real
the form of a multi-layered structure in which the high-level vehicle (in the case of eventual experiments on the real car).
controller realizes the blended ABS logic, and the low-level The actual wheel velocity Vw and longitudinal vehicle
fuzzy controller determines the hydraulic actuation of the acceleration ax serve also as the input for the slip and velocity
system. The high-level controller comprises feedforward and observers.
feedback compensation elements.
©

The main blocks of the ABS are further introduced in more

The analysis of the mentioned and other publications allowed details (realization of friction, slip and velocity observers is not
to point out that many developed methods of electric motor discussed within the framework of this paper).
control in ABS mode have been validated only with computer
904 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014)

Generation of Reference Slip

The reference slip generator is designed to determine the
reference slip λref by means of a tyre model. On the first step
the reference slip is defined as

(3)

l
The reference slip generator is realized in the form of a lookup

na
table. A graphical interpretation of LUT for the target vehicle
configuration is given in Fig. 3.

io
at
Figure 2. General structure of ABS controller

Torque Demand
The generation of the torque demand Tdem is calculated from
the driver control action, in particular from the brake pedal
travel. In the presented study, the practical ABS realization
includes a decoupled electro-hydraulic brake system with the
embedded pedal unit that is able to provide the corresponding
rn Figure 3. Generation of reference slip

On the final stage a slip target adaptation (STA) algorithm is

te
signal of the pedal position [5]. Preliminary the value of applied to modify progressively the reference value obtained
Tdem_prim is processed as function of the brake pedal travel from the LUT, in order to be as close as possible to the
through a lookup table (LUT). Then, this value is corrected maximum braking torque and to avoid under or overbraking
caused either by inaccurate estimations of μmax or by tire-road
In

through the predictive torque in accordance with the road

conditions to prevent vehicle underbraking / overbraking. behavior different to the one modeled by the LUT. The
adaptation is realized through the correction of the reference
The predictive torque is defined as: slip in accordance with the estimation of the sign of the actual
slope of the µx-λ curve. As the road friction coefficient µx is
directly related to the longitudinal acceleration ax, the
E

estimation of the road µx-λ slope can be done as follows:

(1)
SA

where μmax_est is the estimated value of maximum friction

coefficient for given road conditions, Fz is the estimated normal
wheel load, r is the tire rolling radius, kpred is the correction (5)
coefficient. It is determined for the target vehicle that kpred=0.25
for front wheels and kpred=0.5 for rear wheels.
Fig. 4 shows the adaptation procedure of STA algorithm.
After the processing of the predictive torque, the torque
requested by the driver is saturated as:
©

(2) Figure 4. Adaptation procedure of STA algorithm

Positive sign indicates underbraking, whilst negative sign

would mean that the maximum friction peak is overcome, so
the slip reference has to be reduced. The adaptation law is
based in the integration of the sign estimation times a gain k.
 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014) 905

For the real implementation, adaptation is only enabled under Torque Blending
specific conditions: when the sign estimation is considered The blending between Tem_dem and Tbr_dem has to fulfil three
significant in terms of numerical stability and coherency objectives: maximize energy regeneration, perform high
between wheels, and when the slip tracking is accurate frequency modulation of reactive torque in ABS situations
enough. taking advantage of the electric motors fast response, and do
not overcome motor and battery regeneration limits for low
Reactive Torque Controller frequency demands. The algorithm has to reach these
objectives under three main working conditions:
The general control law for reactive torque is described as:

l
1. Total torque demand (Tdem) in the four wheels is below

na
the maximum regenerative torque available because of
(6) battery state and individual torque demands are below
available torque at corresponding motors: In this case, the
where the correction factor ξdriver_dem is used for the saturation brake blending allocates all torque demand to the electric
of the reactive torque in order to track the driver demand and motors.
to prevent generation of wrong torques. The factor ξdriver_dem is

io
2. No regenerative torque is available (e.g. full battery): The
calculated depending on the driver demand.
friction brakes have to provide a brake torque level, and
the electric powertrain generates the modulation over
The main component of the reactive torque controller is the
and under this level by applying alternatively braking and

at
proportional-integral (PI) part realising the direct wheel
traction torque so that the energy regeneration is null or
slip-based control. The PI controller forms the control demand
slightly negative. To do that, the algorithm implements a
νPI:
filter for the reactive torque Treact. Then, the low frequency
components of Treact and Tdem are assigned to the friction
rn (7)
brake torque Tbr_dem and the high frequency components
are assigned to the electric motor torque Tem_dem.
3. Intermediate cases: In the rest of cases, the electric
powertrain has to modulate the fast torque variations
te
The proportional part of the controller is the function of the whilst the brake torque “slow” level is split between friction
vehicle velocity brakes and regeneration, taking into account motor
and battery limitations but giving priority to regenerative
brakes.
In

(8)
It should be noted that the proposed architecture has an
and the integral part is computed as: external active vibration controller to reduce motor torque
oscillations caused by powertrain dynamics, based on the
feedback of motor and wheel speeds. Active vibration control
E

algorithm is described in [10].

SA

(9) Procedure of Definition of Controller Gains

The following criteria are used to find proportional and integral
where the coefficient α is the modifying factor to define the gains in reactive torque controller in the first approximation:
quickness of integral part changing.
1. Recursive mean square error of longitudinal slip
In Eqs. (8) and (9), the variable e is the control error. The
saturation of the control error is effected in the following way:
©

(11)

(10) 2. Integral of time-weighted-absolute-error of longitudinal slip

(12)
906 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014)

3. Total variance of control demand VALIDATION OF DEVELOPED ABS

CONTROLLER
Target Vehicle Configuration
The results presented in the actual study refer to the sport
(13) utility vehicle with technical data given in Table 1. The vehicle
The resulting cost function for the gain optimization can be has the powertrain configuration with four on-board electric
proposed from the criteria 1)-3) as follows: motors, gearboxes, and half-shafts between the drive motors
and the wheels, Fig. 6.

l
Table 1. Technical data of the vehicle

na
(14)

io
The next step is the gain scheduling dependent on vehicle
velocity. For this purpose the step braking maneuver from
different initial velocities has been analyzed by means of

at
the Differential Evolution optimization strategy The detailed
discussion of the procedures of definition and optimization
of controller gains is out of scope of this paper. However, for
illustrative purposes, Fig. 5 demonstrates the variant of gain
scheduling for the developed reactive torque controller that
has been used in subsequent ABS tests. As it can be seen, the
proportional gain is being decreased and the integral gain is
be increased with the brake velocity degradation. It is required
rn
te
to guarantee low oscillatory behavior of actual slip near to the
reference value as well as adaptation of ABS frequency. The
gain gradients are distinctly changed after 50 km/h to allow
smooth transition of ABS operation from slip-based to velocity-
In

based control.
E

Figure 6. Powertrain configuration for one axle [11]

SA

Friction brakes are actuated through decoupled electro-

hydraulic system (Slip Control Booster (SCB) development of
TRW Automotive company), Fig. 7.
©

Figure 5. Gains of reactive torque controller depending on vehicle

velocity (proportional gain is scaled down on 103; integral gain is
scaled down on 104) Figure 7. Electro-hydraulic brake system [5]
 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014) 907

Experimental Setup Input is a sine sweep, with offset 100 Nm and amplitude 10 Nm
Within the framework of the presented paper, the results of at 4 Hz/s, to the respective controllers and the half shaft and
validation of developed ABS controller on real-time hardware- SCB unit torque are the outputs. From Fig. 9, which shows the
in-the-loop (HIL) test rig are discussed. Fig. 8 shows the main results, it can be seen that the electric driveline and SCB unit
components of the experimental setup [12]. It is based on have a −3 dB bandwith of 14.7 Hz and 9.6 Hz respectively.
dSPACE modular platform with several components Due to the non linearity of the actuators, the bandwith is slightly
responsible for data input/output, the control on brake system dependant of the torque demand (This dependency is not
and communication with the commercial vehicle simulator IPG shown in the presented paper).
CarMaker.

l
These tests confirm the fact that an improved ABS
performance should be possible in case of blended actuation

na
of the friction brakes and electric drivetrain, as compared with
commonly-used hydraulic ABS, where modulation is usually
around 9 Hz.

io
at
Figure 8. Components of experimental setup
rn
te
The friction brake part of the HIL test rig is represented by an
electro-hydraulic brake system (EHB). A set of sensors
measure brake pressure in each brake caliper. The brake
pedal is actuated by the hydraulic brake robot providing the
In

maximum pedal force up to 1500 N and pre-defined pedal Figure 9. Frequency response of the electric drivetrain and SCB unit
velocity in the range from 0 to 1000 mm/s. with sine sweep input (with offset 100 Nm and amplitude 10 Nm at 4
Hz/s)

Real-time emulators of electric motors and powertrain

components have software realization. They are parameterized Test Program
E

through experimental data obtained by the motor producer The presented paper is concentrated on the analysis of test
INVERTO. results for four case studies:
SA

The full software simulator of the target vehicle is developed in • ABS braking from 60, 90, and 120 km/h on low (μmax=0.3),
the IPG CarMaker software environment. The road conditions, middle (μmax=0.6) and high (μmax=1.0) friction road. No slip
profile of vehicle maneuvers, and the driver model are also target adaptation, the system uses constant reference
tuned through IPG CarMaker program shell. slip. This maneuver is used for the analysis of the system
behavior at different constant reference slip ratios.
Electric Motors and Slip Control Booster • ABS braking from 120 km/h on low (μmax=0.3), middle
(μmax=0.6) and high (μmax=1.0) friction road. Slip target
Performance
adaptation is activated.
©

Simulation models, which have been validated on the HIL rig

and vehicle demonstrator, are used to carry out actuator • Comparative ABS braking from 120 km/h on low
performance tests. (μmax=0.3) and middle (μmax=0.6) friction road for two
different values of torsional stiffness of powertrain: 9000
and 21000 Nm/rad.
• Emulation of failure of the vehicle control unit.

The corresponding results are discussed in next section.

908 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014)

ANALYSIS OF CASE STUDIES

1) ABS Braking without Slip Target Adaptation
For the case of deactivated adaptation of the target slip value,
four variants of constant reference slip (λref = 0.04, 0.1, 0.15,
and 0.25) were tested for all claimed braking conditions. Fig.
10, 11, 12 show examples of corresponding brake diagrams for
several particular test cases. After analysis of the experimental
data, the following conclusions can be done.

l
na
1.1. Tracking of reference slip. From Fig. 10, 11, 12, 13 can be
concluded that the developed ABS shows smooth tracking
control during steady state condition and robustness
regarding the relevant range of reference slip values at Figure 10. (cont.) Velocity profile and distributions of torques on front
different velocities on low- and middle-μ roads. On low-μ left and rear left wheels by braking from 120 km/h on low friction road;
reference slip 0.1

io
road, irrespective of the velocity at braking, the controller
starts with initial phase of slip adaptation characterized
by very small drop of the slip ratio (see example on Fig.
13). In the worst case (120 km/h), this drop is of 0.11

at
higher as the reference slip for λref = 0.25. These initial
drops are of very short duration. It should be pointed
out that the system behaviour is smoother in the case
of ABS with activated predictive torque block. Below 30
km/h the reference speed is tracked, instead of reference
rn
slip, due to an increased influence of the noise in the slip
estimation, which would cause unstable tracking control.
te
In
E
SA

Figure 11. Velocity profile and distributions of torques on front left and
©

rear left wheels by braking from 120 km/h on middle friction road;
reference slip 0.1

Figure 10.
 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014) 909

l
na
Figure 11. (cont.) Velocity profile and distributions of torques on front Figure 12. (cont.) Velocity profile and distributions of torques on front
left and rear left wheels by braking from 120 km/h on middle friction left and rear left wheels by braking from 120 km/h on high-friction road;

io
road; reference slip 0.1 reference slip 0.1

1.2).Brake distance. Fig. 14. ABS controller ensures the 1.3).Torque distribution and blending. The specific character
reduction of the brake distance on low-μ and middle-μ of the performed experiments has allowed to evaluate

at
road surfaces. In particular, for the given range of the the torque blending during ABS operation with hardware
initial braking velocities, the achieved reduction was from friction brake system and real-time emulated electric
5.4% to 8.3% on the middle-μ road and from 14% to powertrain in a braking mode. In all the tested cases,
16.8% on the low-μ road as compared with the braking the employment of the friction brake system has been
without ABS. The influence of ABS operation on the
brake distance on the high-μ road can be considered as
not significant, within ± 2%. Lower reference slip ratios
provide reduced distances.
rn minimized to increase the regenerative braking by electric
powertrain fulfilling required total brake torque. The ABS
has practically minimized the operation of friction brake
system at braking on low-μ road. On the high-μ road, the
te
ABS has some torque oscillations of low amplitude with
frequency of 1…3 Hz depending on the vehicle velocity.
In
E
SA
©

Figure 13. Wheel slip profiles on front left and rear left wheels by
braking from 120 km/h on middle-friction road; variations of constant
Figure 12. reference slip
910 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014)

2) ABS Braking with Slip Target Adaptation

In this case study, target slip values are stored in look-up
tables (LUT) generated as functions of the normal wheel force
and the tire-road friction characteristics. Fig. 15 displays
diagrams illustrating the effect of the slip adaptation on three
different road surfaces.

Analysis of the results leads to the following conclusions.

l
2.1. Brake distance. The tests of ABS with activated slip target

na
adaptation showed a brake distance reduction in all the
cases in comparison to the tests where the STA block was
Figure 14. (cont.) Brake distances for ABS braking with different
deactivated, Table 2.
reference slips

Table 2. Comparison of brake distances for braking from 120 km/h 2.2.Tracking of reference slip. In all the tested configurations,

io
the system has reached the reference slip ratio within
1 sec after the start of the braking process. The single
insignificant deviations from the reference slip are
observed on middle- and high-friction road in the middle

at
part of the braking process.
2.3. Slip target adaptation results in progressive variation of
the initial reference slip target. For low mu surfaces, there
is a clear reduction of target with regard to the LUT output,
rn both in front and rear wheels.

Typical increase of slip values has taken place for all road
surfaces to the end of manoeuvre. An important fact is that no
te
wheel blocking has been also observed at low velocities (<20
km/h). The final slip values did not surpassed 0.2 on low-μ
road, 0.5 on middle-μ road, and 0.65 on high-μ road.
In
E
SA

Figure 15. Wheel slip profiles for ABS braking with slip target
©

adaptation on different road surfaces

Figure 14.
 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014) 911

l
na
Figure 16. (cont.) Wheel slip profiles for ABS from 120 km/h,
powertrain stiffness 9000 Nm/rad

io
The results have displayed no deterioration of brake
performance in terms of brake distance and average
deceleration. The slip dynamics of the ABS with reduced

at
powertrain stiffness is also characterized with sufficiently
smooth tracking of reference slip ratio both on front and rear
values. These results are confirmed that the developed ABS
control strategy can be used in the case of change of the
rn powertrain configuration, in particular, for less stiff variant of the
half-shafts between electric drive motors and the wheels.
te
4) ABS Failure
Figure 15. (cont.) Wheel slip profiles for ABS braking with slip target
A special test procedure relates to the emulation of the failure
adaptation on different road surfaces
of the vehicle control unit. In such situation, the internal ABS
3) Comparative Test of ABS Braking for controller of the SCB system should be activated to complete
In

Different Powertrain Parameters the braking maneuver. Using the real-time test on the HIL test
platform, the corresponding braking test has been performed
To investigate a possible influence of powertrain parameters on
for the worst-case scenario with initial velocity 120 km/h and
the ABS performance, additional experiments have been done
low-μ road. As it can be seen from Fig. 17, the failure of the
for the vehicle parameters with reduced stiffness of the
ABS occurred after 8 s of the braking process. The controller
E

powertrain of 9000 Nm/rad instead of initially used stiffness of

has recognized the critical situation and activated internal ABS
21000 Nm/rad. The selected wheel slip profiles at ABS braking
of the SCB system within 200 ms. For the given conditions, a
with reduced stiffness are given in Fig. 16. Intensive slip
short-term increase of the slip took place without the locking of
oscillations at the end of braking maneuver (by velocity below
SA

wheels. Hence, the coordinated operation of the developed

30 km/h) are caused by changing of the control strategy
ABS and internal ABS of SCB unit is confirmed.
(velocity control instead of slip control) and are subjected to
more precise tuning of control gains.
©

Figure 17. Braking with ABS failure

Figure 16.
912 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014)

SUMMARY 3. Freitag, G., Gerlich, M., Bergmann, D., Pais, G. and Fischer,
B., “Replacement of the Friction Brake by a Wheel Hub Drive,”
The presented study has introduced a new ABS control presented at the 3rd International Munich Chassis Symposium
chassi.tech plus, Munich, Germany, June 21-22, 2012.
strategy developed for electric vehicles with individually
4. Bottiglione, F., Sorniotti, A., and Shead A., “The Effect of Half-
controlled on-board motors for each wheel. The controller is Shaft Torsion Dynamics on the Performance of a Traction
based on PI direct slip control and has been experimentally Control System for Electric Vehicles,” Proc. of the Institution of
Mechanical Engineers, Part D: Journal of Automobile Engineering.
validated and verified using hardware-in-the-loop technique. A 226 (9): 1145-1159, 2012, doi:10.1177/0954407012440526.
number of conclusions can be deduced from the ABS 5. Savitski, D., Ivanov, V., Heidrich, L., Augsburg, K. and Pütz,
validation for straight-line braking maneuvers in different road T., “Experimental Investigation of Braking Dynamics of Electric
Vehicle,” presented at Eurobrake 2013, Dresden, Germany, June
conditions: 17-19, 2013.

l
6. Song, C., Wang, J. and Jin, L., “Study on the Composite ABS

na
• The proposed control strategy for ABS actuating electric Control of Vehicles with Four Electric Wheels,” Journal of
Computers, 6(3): 618-626, 2011, doi:10.4304/jcp.6.3.618-626.
motors in a barking mode makes it possible to achieve
7. Bera, T.K., Bhattacharya, K. and Samantaray, A.K., “Bond
smooth tracking of reference wheel slip as compared with Graph Model-Based Evaluation of a Sliding Mode Controller for
conventional hydraulic ABS. a Combined Regenerative and Antilock Braking System,” Proc.
of the Institution of Mechanical Engineers, Part I: Journal of
• No essential oscillations of electric motor brake torque Systems and Control Engineering, 226(8): 1060-1076, 2012,

io
have been observed during the ABS actuation. The doi:10.1177/2041304110394558.
highest frequency of torque oscillation, 1…3 Hz, has 8. Zhang, J.L., Yin, C.L. and Zhang, J.W., “Improvement of
Drivability and Fuel Economy with a Hybrid Antiskid Braking
taken place for the test cases on high-friction road System in Hybrid Electric Vehicles,” International Journal of
surfaces. However the amplitude of these oscillations was Automotive Technology, 11(2): 205-213, 2010, doi:10.1007/

at
s12239-010-0026-0.
insignificant.
9. Sado, H., Sakai, S. and Hori, Y., “Road Condition Estimation
• In the case of ABS without slip target adaptation, the for Traction Control in Electric Vehicle,” presented at IEEE
reduction of brake distance as compared with the braking International Symposium on Industrial Electronics ISIE, Bled,
Slovenia, July 12-16, 1999, doi:10.1109/ISIE.1999.798747.
without ABS was insignificant for high-friction road
surface. For low- and middle-road surfaces, the achieved
brake distance reduction was in the range 5.4…16.8%
depending on initial braking conditions and appointed
rn 10. Rodríguez, J.M., Meneses, R., Orús, J., “Active Vibration Control
for Electric Vehicle Compliant Drivetrains,” presented at IECON
2013 Conference of the IEEE Industrial Electronics Society,
Vienna, Austria, November 10-13, 2013.
11. Dhaens, M., “Recuperative Braking in Electric Vehicles: Flemish
te
reference slip value. and European Initiatives,” presented at the IQPC 4th Intl.
Congress Electric Vehicles, Berlin, Germany, December 5-7,
• The proposed slip target adaptation allows to reach the 2011.
required reference slip value within 1 sec after the start of 12. Heidrich, L., Shyrokau, B., Savitski, D., Ivanov, V., Augsburg,
braking maneuver. No wheel blocking has been occurred K. and Wang, D., “Hardware-in-the-Loop test Rig for Integrated
Vehicle Control Systems, “presented at the 7th IFAC Symposium
In

during the adaptation process. on Advances in Automotive Control, Tokyo, Japan, September
• The developed controller ensures stable ABS operation 4-7, 2013, doi:10.3182/20130904-4-JP-2042.00027.
for powertrain configurations with different stiffness. In
particular, required tracking of the reference slip has been
CONTACT INFORMATION
also achieved for more critical case with low stiffness 9000
E

Nm/rad, where significant powertrain oscillations were Valentin Ivanov

initially expected. Department of Automotive Engineering
Ilmenau University of Technology
• The proposed fail safe mechanism guarantees the
valentin.ivanov@tu-ilmenau.de
SA

switching of the ABS operation to the internal controller

of SCB electro-hydraulic brake system in the case of
the failure of the main ABS controller. The duration of ACKNOWLEDGMENTS
the switching process does not exceed 200 ms, and the The research leading to these results has received funding
wheels are not locking during the corresponding transient from the European Union Seventh Framework Program
process. FP7/2007-2013 under grant agreement no. 284708

The authors plan to introduce results of ABS testing on real

©

vehicle in further publications as next stage of the presented DEFINITIONS/ABBREVIATIONS

research. ABS - anti-lock brake system
FL - front left wheel
REFERENCES FR - front right wheel
1. Akaho, D., Nakatsu, M., Katsuyama, E., Takakuwa, K. and
Yoshizue, K., “Development of Vehicle Dynamics Control System HIL - hardware-in-the-loop
for In-Wheel-Motor Vehicle,” presented at JSAE Annual Congress RL - rear left wheel
(Spring), Yokohama, Japan, May 19-21, 2010.
2. Murata, S., “Innovation by In-Wheel-Motor Drive Unit,”. Vehicle RR - rear right wheel
System Dynamics. 50 (6): 807-830, 2012, doi:10.1080/00423114.
2012.666354. SCB - Slip Control Boost
 Ivanov et al / SAE Int. J. Passeng. Cars - Mech. Syst. / Volume 7, Issue 2 (August 2014) 913

STA - slip target adaptation

e - control error
Fdriver - driver control action
Fem - brake force generated by electric motor
Fz - vertical tyre force
ITAE - integral of time-weighted absolute error
k - actuator rate
KP - proportional gain

l
KI - integral gain

na
LUT - Look-up table
r - tire radius
RMSEλ - recursive mean square error of longitudinal slip
s - Laplace operator

io
sdist - brake distance
Tbr - friction brake torque
Tbr_dem - friction brake torque demand

at
Tdem - torque demand
Tem - electric motor torque
Tem_dem - electric motor torque demand
Tpred - predictive torque
Treact - reactive torque
Tw - wheel torque
rn
te
TV - total variance
u - control demand
Vw - linear wheel velocity
In

Vx - longitudinal vehicle velocity

λ - wheel slip
λref - reference wheel slip
μ - tire friction coefficient (general)
E

μmax - maximum tire friction coefficient

μx - longitudinal tyre friction coefficient
νdem - correction gain taking into account torque demand
SA

νPI - proportional-integral gain

ξdriver_dem - correction factor of driver demand
©

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical,
photocopying, recording, or otherwise, without the prior written permission of SAE International.

Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE International. The author is solely responsible for the content of the
paper.