International Journal of Crashworthiness

ISSN: 1358-8265 (Print) 1754-2111 (Online) Journal homepage: http://www.tandfonline.com/loi/tcrs20

A new pre-crash system for side impact protection

J. Tandler , E. Zimmerman , V. Muntean , B. Seipel , T. Koch , D. Willersinn , M.

Grinberg , C. Mayer & M. Diez

To cite this article: J. Tandler , E. Zimmerman , V. Muntean , B. Seipel , T. Koch , D. Willersinn ,

M. Grinberg , C. Mayer & M. Diez (2008) A new pre-crash system for side impact protection,
International Journal of Crashworthiness, 13:6, 679-692, DOI: 10.1080/13588260802411499

To link to this article: http://dx.doi.org/10.1080/13588260802411499

Published online: 05 Dec 2008.

Submit your article to this journal

Article views: 117

View related articles

Citing articles: 4 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=tcrs20

Download by: [Indian Institute of Technology Roorkee] Date: 21 September 2016, At: 23:28
International Journal of Crashworthiness
Vol. 13, No. 6, December 2008, 679692

A new pre-crash system for side impact protection

J. Tandler , E. Zimmerman , V. Munteanb , B. Seipelc , T. Kochc , D. Willersinnd , M. Grinbergd , C. Mayere and M. Diezf
a b

a
Continental, Lindau, Germany; b Faurecia, Scheuerfeld, Germany; c Fraunhofer Institute LBF, Darmstadt, Germany; d Fraunhofer
Institute IITB, Karlsruhe, Germany; e Daimler, Boblingen, Germany; f Cidaut, Valladolid, Spain
(Received 1 April 2008; nal version received 14 August 2008)

Sub-project 6 of the European Integrated Project APROSYS, implemented within the 6th Framework Programme, has
developed an integrated safety system for side impact protection. For this, two different innovative technologies have been
applied for the rst time as a technology showcase in a car: shape memory alloy based actuators and a pre-crash sensing
subsystem using radar and stereo-vision observing the side of a vehicle. Several technical challenges had to be met during
development work. The sensing subsystem detects an imminent collision and decides about the probability of a collision.
It activates the actuator before the crash occurs in order to reduce intrusion into the passenger compartment. This article
describes the sensing subsystem, the actuator subsystem and nally the testing programme that was executed to evaluate the
complete system.
Keywords: integrated safety; side pre-crash; shape memory alloy; radar; stereo video

1. Introduction and the relevance of side impact The sensor subsystem consists of the combination of 24-
protection GHz radar sensors with a stereo video camera. The actuator
An increasing number of cars are equipped nowadays with uses shape memory alloys (SMAs) to realise a fast but
intelligent safety systems. These use information from en- reversible actuator. The system was kept generic, so that it
vironmental sensors to detect possible threats and can adap- could be built into any vehicle.
tively take appropriate countermeasures. In particular, pas- As the nal part of the APROSYS SP6 testing pro-
sive safety measures improving the crashworthiness of the gramme, this system has shown its effectiveness in a
car can be deployed in a way that they offer an optimal EuroNCAP-based crash test. Demonstrating the integrated
protection for the car occupants. This approach is widely character of the system, the sensors detected the approach-
known as integrated safety. The integrated safety system ing crash barrier, and 231 ms before the crash triggered the
described here has the objective to mitigate injuries in side actuator. As a crash facility is an unnatural environment for
crashes. a sensing system and one single test is not enough, the sen-
Side crashes are a major threat to trafc safety. While sor system evaluation took place beforehand. The actuator
front and side crashes occur with comparable frequency, the reduced intrusion and intrusion velocity into the passenger
risk of an injury from a side crash is much higher. This is due compartment. This gives conventional devices, like airbags,
to the close distance between an occupant and the incoming more time and space to protect the car occupants.
object at rst contact. For the same reason, existing in-crash
sensing technology does not allow for timely deployment of Detecting imminent side crashes: the sensor
collision mitigation measures. In order to develop a suitable subsystem
side impact protection system in APROSYS sub-project 6 From a detailed study of accident statistics, requirements for
(SP6), a complete technical development programme was the sensing subsystem were derived [11]. In order to detect
executed. Firstly, accident statistics were analysed in detail. a large number of side impacts and provide useful informa-
On the basis of this, various concepts for side impact pro- tion to the actuator subsystem, the sensing subsystem has to
tection were studied in multi-body [1] and nite elements
simulations [10]. Also human behaviour in side crashes was r detect and track objects impacting the side of the car;
examined [7]. Accordingly, the sensor and actuator subsys- r determine size and shape of impacting objects (for
tems were dened, developed and integrated into several classication: pole motorcycle, car);
vehicles. A thorough evaluation programme has tested sen- r determine the collision risk and decide about a colli-
sor and actuator subsystem. sion at the latest 200 ms before impact;

Corresponding author. Email: joachim.tandler@continental-corporation.com

ISSN: 1358-8265
Copyright 
C 2008 Taylor & Francis
DOI: 10.1080/13588260802411499
http://www.informaworld.com
680 J. Tandler et al.

Figure 1. Field of view of the sensing system (radar sensors, blue; stereo camera, red).

r survey an area of 20 m range and the side/front of In a front view, vehicles are mostly seen from the front
the car. or from the rear, and they are usually at such a distance
that they are completely visible in the image. These restric-
This is achieved by a fusion system consisting of tions make it possible to exploit generic features such as
r a radar sensor network (two sensors) and symmetry and U-shapes for object detection in real time.
r a stereo video camera. Another particularity of the front view is that observed
vehicles mainly move either parallel or anti-parallel to the
camera-carrying vehicle. As a consequence, vehicle orien-
The mounting positions and the eld of view of these
tation may be assumed constant which allows for the use of
sensors are indicated in Figure 1.
simplied motion models for vehicle tracking. Also, track
In the rst step, the information from both sensor types
initialisation is considerably simplied if the vehicle ori-
is processed individually. Then, a fusion module combines
entation can be assumed to be known. Detection of lane
their data before it is used by a decision module to assess the
markings allows nding regions of interest and makes it
collision risk and eventually trigger the actuator subsystem.
additionally easier to identify relevant objects.
The restrictions of a front view do not apply in a side-
The radar sensor network looking perspective. As Figure 3 illustrates, many different
object orientations have to be taken into account. Combined
The Continental short range radar was adapted for the spe-
cic needs of the side-looking application. Its key features
are:
r Centre frequency 24 GHz
r Signal bandwidth 1000 MHz
r Measures simultaneously:
- Distance (accuracy 0.1 m)
- Angle (accuracy 2 )
- Relative speed (accuracy 1 m/s)

For smooth integration under front and rear bumpers,

the radar antenna was adapted to look into front/side direc-
tion (Figure 2).

The stereo video camera

Together with the radar network, a stereo video camera
system was chosen in APROSYS as a sensor for pre-crash
detection. High resolution combined with decreasing cost
make vision-based systems more and more attractive for
automotive applications. While developing a side-looking
Figure 2. Beam patterns of the adapted radar sensor prototype.
video pre-crash system several particularities have to be Although the sensors can be easily mounted parallel to the cars
taken into account, which make such a system different symmetry plane, their sensitivity is concentrated in the front/side
from systems used for forward-looking applications. direction as required.
 International Journal of Crashworthiness 681

After extraction of a depth map, road location is esti-

mated. Then, object points are separated from road points
and grouped into clusters. Tracking of those clusters makes
it possible to estimate motion of observed objects and to pre-
dict their behaviour in advance. Together with ego-motion
estimation this allows for collision prediction.
To cover the observation area shown in Figure 1, the
individual cameras in APROSYS have wide eld of view
lenses and a non-standard orientation with respect to the rig
base. The camera orientation is depicted in Figure 4.
This leads to some particularities in low-level stereo
data processing. In standard stereo systems, the input
Figure 3. Scenarios with complex object trajectories observed images are rst transformed to epipolar geometry. The
by a side-looking video camera. advantage is that corresponding features are located in
the same image line in both images. If the cameras
are oriented as shown in Figure 4, then the images in
epipolar geometry show large trapezoid image distor-
tion, cf. Figure 5. The transformation would make time-
consuming and noise-prone interpolation necessary. For
this reason we skip the transformation to epipolar ge-
ometry and execute correspondence search directly in
the original images. Corresponding features are then lo-
cated on complex curves rather than on straight lines, see
Figure 6.
Figure 4. Non-standard camera orientation in APROSYS. Further challenges of a side-looking video-based pre-
crash system are small observation time and complex object
trajectories, combined with relatively low sampling rate of
with a large variety of object types, this makes it extremely the video system (compared e.g. to a radar or lidar) and
challenging to detect objects based on their appearance in a long period for which prediction has to be done. This
real time. For the realisation of a side-looking video system implies usage of a tracker with appropriate object motion
we therefore had to choose a generic approach to achieve models [4].
robust and reliable object detection. Object hypotheses are In Figure 7, results of side pre-crash detection based on
generated from so called depth maps which are the result of stereo processing are shown. On the left side, the input im-
stereo processing. Depth maps consist of three-dimensional age with depth measures and their projection to the ground
point clouds. They are built by nding corresponding struc- plane is shown. On the right side, a top view illustrating the
tures in both video images and reconstructing their range result of processing is displayed. Dots represent depth mea-
using knowledge of mutual orientation of both cameras. sures; box represents the object hypothesis. The estimated

Figure 5. Visualisation of epipolar lines in rectied images.

682 J. Tandler et al.

Figure 6. Epipolar curves (magenta) for correspondence search in original image geometry.

object position, relative speed and time to collision (TTC) The combination of both sensor types makes it possible
are printed in red. to suppress signicantly false alarms while maintaining a
high efciency to detect side impacts.

The fusion system The decision module

The fusion system enables spatial alignment and synchro- The decision module performs a risk assessment: it extrap-
nised data acquisition of the radar and the video subsystem. olates all detected objects in time and calculates eventually
This is crucial for the demanding side pre-crash applica- the TTC. It determines if it is physically possible to escape
tion. The fusion system can associate matching video and the collision by braking or steering and calculates accord-
radar objects and provide them to the decision module in a ingly a collision probability. If TTC and collision probabil-
consistent format (cf. Figure 8). ity cross certain thresholds, the decision that a collision will

Figure 7. Result of side pre-crash detection based on stereo video.

 International Journal of Crashworthiness 683

to react adaptively and take optimal countermeasures with

an appropriate actuator.

Adaptively improving the crashworthiness of a car:

the actuator subsystem
In side impacts, the fatalities normally occur within the
rst period of the crash (7080 ms) and are mainly caused
by the hard contact of the occupant with the interior trims
when the biomechanical limits for the head deceleration
and thorax loads are exceeded. However, severe injuries can
occur at the pelvis and lower extremities if the contact with
the interior trim is too hard. On the top, the small energy-
absorbing zone available on the side structure limits the
protection possibilities.
Looking closer at side impact accidents with respect to
Figure 8. Schematic view of radar and video information to be structural deformation, three very different crash scenarios
fused. can be distinguished. These are car to car, car to pole
type road infrastructure and car to motorcycle impact.
To ensure protection in all these scenarios a safety system
happen is taken and communicated to the actuator system must cope with the specic protection needs of these very
(cf. also [5]). different load cases.
An example of a situation when a trigger has to be To reduce the severity of a side impact in the above men-
issued is depicted in Figure 9. tioned scenarios, a two-fold strategy is necessary: reduce
Since the sensor information about the accident cong- the structural deformation of the car body in regions where
uration is thus available before the crash, it is still possible the occupant life is jeopardised and secondly soften the con-
tact of the occupant with the interior trim. For APROSYS
SP6 however, it was decided to work on intrusion reduction
only, to focus on devices which try to reduce the deforma-
tion of the car structure.
The actuator subsystem was evaluated in side-crash
tests with a deformable barrier, for which an extensive anal-
ysis of the structural behaviour was performed.

Structural deformation in side impact with

deformable barrier
As a reference, a full car crash according to the EuroNCAP
protocol was performed, which led to a good understand-
ing of the structural crash modes in side crash. Intrusion
measurements were done by acceleration sensors mounted
at signicant locations on the B-pillar, rocker and door
panel on both car sides as well as post-crash intrusion mea-
surements. As an outcome of this study, relevant structures
driving deformation (B-pillar, rocker, roof, oorboard and
doors) were identied and analysed.
From there, the B-pillar is considered to be the structural
Figure 9. Top view of a scene representation in the fusion and part driving the intrusion in side crash. The B-pillar does
decision application (unit: meters). The host car carrying the sen- not follow a continuous bending line, but, by its plastic
sors is indicated as a grey rectangle around (0,0). The red rectangle hinges, determines the deformation pattern of the complete
represents a stereo video object. The small white box represents a side structure. Five plastic hinges are developing along its
radar reection object. The lines point to where an object would lengths (Figure 10). It can be observed that the hinge at the
be after 1 s with linear extrapolation. Red boxes indicate possible
impact points on the host car surface. In this case, a trigger has to height of the belt line intrudes the most into the passenger
be sent to the actuator system. compartment (488 mm).
684 J. Tandler et al.

Figure 10. B-pillar hinges in side crash with MDB.

Actuator system layout Generally, the actuator system comprises of three re-
Different concepts to reduce intrusion were investigated. gions as depicted in Figure 11:
They consisted of either connecting the doors to the car r Door seat coupling (corresponding to struck side)
body in different congurations, or supporting B-pillar or
redirecting crash loads towards stiff regions, especially non- done by an active device, an SMA tube actuator
r Seat-centre console-seat coupling
impacted structures like B-pillar, oorboard and crash box r tambour spring pre-dened rotation hinge located
[6,8]. For the investigation of the broad variety of concepts
to reduce intrusion in side impact, a multi-body model and on the unstruck side at rocker to oorboard connec-
a generic FE model of Chrysler Neon were used. tion
From the investigated concepts it was found that the Broken down into components, the actuator system is
best way to reduce intrusion was to redirect the loads to the composed of a transversal tube (5) (Figure 11), reinforce-
unstruck side, where the rocker meets the crash box and the ments to prevent buckling (6 and 7), additional supports of
oor. This region was considered to be very stiff with a low the beam on the unstruck side (811) and active parts (2
risk of injuries for the occupants. and 4) actuated to close the load mechanism.
The redirection of the loads actively opens a new load The principle of the active device is shown in Figure 12.
path from the rst contact phase on and consequently un- The active device (the actuators) is located on the struck side
loads the B-pillar. On top, it reduces the intrusion not only and consists of two parts. The rst one is a rotating door
at the B-pillar but in regions where the occupant sits. beam which in an un-deployed state stands upright and does

Figure 11. Actuator layout in simulation.

 International Journal of Crashworthiness 685

Figure 12. Active part function.

not hinder the normal travelling of the glass. When red, this actuator system behaviour under crash loads and prevent
beam rotates into a horizontal position lling the complete it from buckling or any other undesired intrusion in the
door box. The second part is a tube with a popping out passenger environment. The tambour spring located on the
bolt which is mounted in the seat, transversally just below unstruck side is meant to take the crash loads from the
the knees. In normal driving conditions the space between barrier and to put them to the car structure by a pre-dened
seat and door is free, when deployed, the bolt creates a tight rotational deformation pattern. In this way, the crash peaks
engagement of the door and the transversal tube mounted in transferred to the complete actuator system are considerably
the seat. In this way, all existing gaps between the outer sheet smoothened. This applies especially for the high amount of
metal and the seat are closed. By this the actuators fulll bending moments on the bolt (Figure 13b).
their safety function to occupy space which is needed for A visual representation from Computer Aided En-
comfort applications during normal driving. Furthermore gineering (CAE) of the simulated intrusion reduction is
components and reinforcements (Figure 11) stabilise the depicted in Figure 14. The intrusion is reduced not only

Figure 13. Actuator failure modes. (a) Crush box rotation on unstruck side; (b) actuator bending on crush side.

Figure 14. Actuator performance. Intrusion reduction at the B-pillar and where the passenger sits.
686 J. Tandler et al.

Figure 15. Schematic representation of the shape memory alloy (SMA) effect. The austenite and martensite states of an SMA material
differ in the congurations of the molecular lattice of the alloy. By heating and cooling one can switch between these two congurations.
Besides, the martensite conguration can be deformed (effect not used in the described actuator).

at the B-pillar region but where the occupant sits as well. one material; in case of SMAs the actuator function is being
In blue, the reference model (no actuator) is shown; the primarily technically exploited. The functional principle of
model with the pre-crash system implemented is shown in this material is based on a thermally induced solid state
red (this gure can be viewed in colour online). phase transformation, from martensite to austenite (Figure
After concept and design of the actuator system were 15). Increasing the temperature refers to the transition of
identied, the nal actuator prototype was realised featuring martensite into austenite. The transition starts at tempera-
an SMA based release mechanism. ture As and is completed at Af (austenite start/nish). While
cooling down the transition is being reversed. The character-
istic temperatures are Ms and Mf (martensite start/nish).
Realisation of a shape memory alloy based tube While the molecular structure of martensite is a face-
actuator centred cubic crystal, the austenite phase shows a body-
Currently smart materials like SMAs or piezoceramics are centred cubic lattice. While heating the alloy, the phase
not in standard use for crash applications. Furthermore, transition from martensite to austenite occurs, the shape of
there was a common understanding that SMAs are not suit- the SMA-device changes and high actuator forces can be
able for time-critical safety applications. Here we describe generated. Feasible shape changes range up to 8% whereas
the development of a prototype actuator, which has an SMA typically active strains are kept limited to roughly 2%. To
release mechanism and is well suitable for the time-critical adjust the working temperature of such SMA actuators,
safety application of APROSYS SP6. First the functional it is essential to design the temperature As . This can be
principle is explained, followed by the mechanical and elec- done in the range of 150 C to 200 C and is achieved by
trical development steps. At the end of this section the nal adjusting the constituents of the alloy. The most prominent
actuator design is shown. More information about the SMA type of SMA is nickeltitanium (NiTi) based. Depending
actuator can be found elsewhere [2]. on operational loads (thermal, strain, mechanical) and type
of alloy such SMA show a lifetime of 106 cycles.
Several possibilities have been checked to heat an SMA-
The principle of shape memory alloy device. Using the car cooling liquid, motor oil or chemicals
Shape memory alloys fall into the category of so called are rejected because of safety risks. So the heating is re-
smart materials. Such smart materials transform energy alised by using the mechanism of the electric conductor
types, e.g. electrical energy into mechanical energy. Of- heating as a result of electricity. This electricity has to be
ten the materials prove both actuator and sensor features in delivered by the car battery. A rst assumption shows that
 International Journal of Crashworthiness 687

Figure 16. Functional principle of tube actuator: (a) initial position; (b) working position.

cable to a Controller Area Network (CAN) and the on board

power supply (Figure 17).

Electrical development
The main part of the electronics is a microcontroller that
has two tasks to provide the communication with the sensor
system via CAN-bus and to control the activation of the
SMA-wire. To activate the SMA-wire it has to be electri-
cally heated in a very short time. This heating is done by a
pre-dened discharging of a capacitor. Instead of one large
capacitor, ve smaller ones (1500 F each) were connected
in parallel. Because of the limitation of the on board power
supply of the car (12 volts), the stored energy of the ca-
pacitors is boosted by using a step up converter to raise
Figure 17. Tube actuator with fully integrated electronics. the voltage up to 40 volts. Due to the high peak current a
special MOS-FET is used for switching the load.
The whole electronics is integrated into the plastic hous-
a normal car battery is able to heat a mass of about 10 g ing by dividing it into three small circuit boards and stacking
within 100 ms. them.

Tube actuator Final tube actuator design

Figure 16 shows a schematic drawing of the tube actuator. The nal design is shown in Figure 18. The tube actu-
A bolt (1), which normally rests in a housing, can be ator can easily be mounted inside the seat structure and
driven by a pre-stressed steel-spring (3) from its initial only needs a cable to connect to the power supply of the
position (inside the seat frame) to a working position. At car and the CAN-bus. If the activation signal is sent via
this working position the top of the bolt is guided by a the CAN-bus, the electronics is heating the SMA-wire by
counterpart at the door trim (4). Small moveable radially discharging the capacitors. The SMA-wire releases the pre-
mounted elements (2) lock the bolt there and prevent an stressed spring and the bolt closes the gap between the seat
unhindered backward motion. and the door trim. The small locking elements prevent an
The release mechanism that initiates the movement of unhindered backward motion. The whole deployment time
the bolt consists of two main parts; a small lever arm that in- from acknowledgement of the activation signal till locking
teracts with the rear end of the bolt (5) and an SMA wire (6) the bolt is about 100 ms.
that during activation pulls the lever arm to release the bolt.

Performance of actuator system

Mechanical development The actuator and the other parts of the actuator system were
To show the potential for serial production, the complete installed in test cars. The actuator system installed in the
release mechanism, the SMA wire, as well as the electronics test car used for nal evaluation is depicted in Figure 19.
were integrated in a polyamide plastic housing. By this it The development process was strongly supported by
was possible to build a stand alone crash actuator. For a CAE simulations. The full car model of the DC Neon
proper operation it only needs to be connected by a small was validated for side impact for both MDB and pole
688 J. Tandler et al.

r Full car testing: Four types of crashes were per-

formed referred to in the following as: reference,
deployed actuator, active actuator 01 and op-
timised actuator. These tests were performed ac-
cording to EuroNCAP side impact protocol with de-
formable barrier. The subjected cars were Chrysler
Neons, model year 9499. The four tests are sum-
marised in Table 1.

Table 1. Summary of tests.

B-Pillar belt Pre-crash actuator

Test vehicle line intrusion system
Figure 18. Bolt (1), mounting ange (2), tube with integrated
sliding bearings (3), steel spring drive (4), stacked electronics (5), Reference 488 mm Not implemented
adaptor (6a), plastic housing (6b). Deployed actuator 412 mm Implemented in
deployed state
Active actuator 01 444 mm Implemented, triggered
by a oorboard
impact scenarios. However validation of the CAE models switch
is a mandatory ongoing process. Therefore, tests to relate Optimised actuator 418 mm Implemented, triggered
by a oorboard
the real performance of the actuator system to the CAE switch
environment were done. The tests can roughly be grouped
in two families:

r Component testing: This was meant to assess the To capture intrusion, pre- and post-geometrical mea-
mechanical performance and the reaction time of the surement of the car body were done. On top, a special crash
SMA actuator. The tests show a proper reaction time measurement system has been installed. On board high-
and a clear repeatability: the actuator moves the bolt speed cameras and several crash accelerometers captured
(400 g) for a travel of 115 mm in 60 ms which dynamically the performance and the deformation modes
is well in the project target of 200 ms, a time span of the car structure.
negotiated with the sensor subsystem. Furthermore, A striking improvement of the actuator system can be
it was shown that the actuator is able to carry the seen at the steering wheel. In Figure 20a, the steering wheel
big amount of crash loads (70 kN) as predicted from of the reference vehicle is severely distorted in compari-
simulation. son with the actuator vehicle (Figure 20b) which is barely
touched by the door; this conrms the reduction of intrusion
all over the door.
A closer look at Figure 21 conrms this signicantly.
The intrusion reduction amounts to more than 50 mm and
is spread all over the door but especially where the occu-
pant sits. This translates in valuable survival space for the
occupants! The reduction of deformation velocity of the
door in critical regions for the occupants (corresponding to
thorax and abdomen) rises up to 2 m/s at 25 ms as depicted
in Figure 22.
On top, due to the usage of the on board measurement
equipment, the function of the actuator system could be
observed during the crash. The performance of the actuator
at 25 ms (a critical time for injuries in side impact and nev-
ertheless for airbag deployment) is depicted in Figure 23.
The door in the reference test is intruding into the
passenger compartment (Figure 23a) whereas, with
actuator, the door is kept away from intruding (Figure 23b).
This space gained in the lower region of the door addresses
directly the pelvis and abdomen of the driver (occupant).
Figure 19. Actuator system implemented in a test car. On the unstruck side, the tambour spring receives the
 International Journal of Crashworthiness 689

Figure 20. Steering wheel deformation.

crash loads and rotates the rocker and the oorboard. The
amount of crash loads transfer is to be seen by the move
of the transversal tube at the side of the seat (Figure 23c,
white marks).
In order to check the actuator subsystem performance
not only in one single crash test scenario, variations of
different scenarios were performed in CAE.
These scenario variations consisted of:
r Impact variation (barrier to car, car to pole and car to
car);
r Impact speed variation (60 km/h, 70 km/h for barrier;
and 35 km/h, 40 km/h for pole)
r Impact angle variation (15 , 30 for barrier; and 75
for pole)

The results are depicted in Figure 24 and show good ro-

Figure 21. Performance of deployed actuator. bustness of the actuator system. The presence of the actuator
reduces the intrusion in all investigated scenarios. However,

Figure 22. Reduction of deformation velocity of the door in the thorax region.
690 J. Tandler et al.

Figure 23. 25 ms: Actuator function in side crash. (a) Tambour effect unstuck side; (b) active components on struck side; (c) reference
door.

r The new load path changes the crash modes of the

side structure.

Evaluating a side pre-crash system: the testing

programme
In the nal phase of the project an extensive test suite was
carried out according to an assessment methodology devel-
oped in APROSYS WP 1.3 [3]. Following this methodol-
ogy, the test program was proposed [9] and performed in
the following three test clusters:
Figure 24. Performance of the actuator system in different crash r Assessment of real-world performance: (false alarm
scenarios; intrusion measured at the driver door in the region of
thorax. rate and comfort issues) by track tests and a eld
test. This test cluster represents the area support-
ing information and is primarily focussing on the
the best performance is achieved in barrier impacts (34%) sensing and decision system.
and car to car (26%). Even though there was no special r Assessment of the pre-crash performance: The test
aligning arrangement done for the actuator system and the conditions are based on the accident scenarios taking
pole impact, the simulations show still an improvement in into account also the system and test lab limitations.
intrusion reduction of more than 10%. Surprisingly, the in- Focus on pre-crash sensing system (radar/stereo cam-
trusion was reduced in the US 75 pole impact scenario by era).
20%. These investigations showed the actuator system in r Assessment of the crash performance: For this, the
multiple impact scenarios, obviously, the concept is robust. EuroNCAP side impact test protocol was selected.
The intrusion and deformation speed reduction at the Focus of the crash performance test is the reversible
door on spread regions is the major gain of the actuator crash actuator application.
system. As a consequence, the actuator system enables to
increase the side impact crashworthiness, such as:
Real-world performance
r Restraint systems have more space and time to deploy A false alarm analysis in real trafc has been carried out in
more efciently (e.g. airbag); a Field Test on European roads (France, Italy, Austria, Ger-
r Contact time is delayed and with this contact forces as many approx. 2000 km). Several alarm situations were
well as the other biomechanical values are reduced; recorded and now clustered whether they are real false
r Intrusion is reduced exactly where the occupant sits, alarms (comfort issue) or acceptable w.r.t. the triggered
survival space and time are gained; safety application. Such eld tests can be recommended as
 International Journal of Crashworthiness 691

Figure 25. Typical situations in the eld test (VUT in front) and example for test situation on test track.

essential within the evaluation of a pre-crash system sup- tests are performed with critical objects in a controlled
plemented by track tests to evaluate the sensor performance environment. The results serve the purpose of the now
in specic trafc situations. Figure 25 shows typical scenes extended technical performance evaluation cluster of
from the eld test and the track test. advanced safety systems.

Pre-crash performance Crash performance

In this test cluster, non-destructive tests, representing The crash test, showing the potential of the reversible actu-
accident scenarios, were carried out (cf. Figure 26). The ator by means of a reduction of the intrusion and intrusion

Figure 26. Non-destructive test in test facility examples of test scenarios representing accident situations.

Figure 27. Test conguration according to the EuroNCAP protocol pre-test installation in crash test environment with test rig to qualify
the system for the nal technology showcase within the project.
692 J. Tandler et al.

velocity, nally allows an overall assessment of the com- nologies in the automotive industry. Continental (Germany)
plete pre-crash side protection system. (A pre-test has been supplied radar sensor and fusion know-how and led the sub-
carried out to ensure that a nal demonstration of the com- project. Fraunhofer-IITB institute (Germany) developed a
plete system as a technology showcase could be arranged stereo camera system. Faurecia (France) provided expertise
as well in a crash test facility environment. In Figure 27, the in automotive interior systems and coordinated the actuator
test conguration and the pre-test installation are shown.) system development. Fraunhofer-LBF institute (Germany)
contributed the shape memory alloy technology. CIDAUT
(Spain) and Instituto Superior Tecnico (Portugal) provided
Conclusions and future steps simulations and crash test analyses. Warsaw University of
In summary, the work described here has shown that it is Technology (Poland) studied driver behaviour in simulated
feasible to build a side impact detection system delivering side crashes. TNO (Netherlands) analysed accident statis-
a trigger at the latest 200 ms before the impact happens. tics, helped with the system denition and performed pre-
Already at this time it is possible to determine that a colli- crash tests. Daimler (Germany) coordinated and also per-
sion cannot be avoided. Although side impact detection is formed tests.
inherently more difcult than e.g. frontal impact detection
and although the research system described here would still
need improvement before becoming a product, it is shown References
that there are no fundamental obstacles. [1] J. Ambrosio, Improved protection of vehicle occupants
The actuator system has been effective to reduce intru- in side impacts by using active structural devices, Pro-
ceedings of the II International Conference on Compu-
sion into the passenger compartment. It is clear now that tational Biomechanics, Lisbon, Portugal, 2005, pp. 989
SMA-based actuators full the requirements of an auto- 1002.
motive time-critical safety application, while offering their [2] R. de Lange, Proposal of generic procedures for an assess-
unique property, i.e. to be fast and reversible at the same ment of pre-crash systems, Public APROSYS Deliverable
time. D6.5.1 (Draft version), Document AP-SP650033, 2007.
[3] A. Eggers and R. de Lange, Proposed evaluation methodol-
In general, modern vehicles will implement an increas- ogy for advanced safety systems, Public APROSYS Deliv-
ing number of active and passive safety features. Hence, erable D1.3.2, Document AP-SP130022, 2006.
there will be a growing demand for an intelligent integra- [4] J. Kasgen and B. Seipel, Konzeptndung, Auslegung und
tion of both. Konstruktion eines Verriegelungsmechanismus auf Ba-
The results from APROSYS SP6 will be used by the sis von Formgedachtnislegierungen zur Optimierung des
Crashverhaltens von Kraftfahrzeugturen, Diploma thesis,
partners in their individual products. We expect that in the Fraunhofer Institute LBF, Darmstadt, Germany, 2004.
coming years we will see more and more vehicles equipped [5] K. Matusiak, T. Dziewoski, Z. Lozia, and M. Guzek, As-
with environmental sensors to obtain pre-crash informa- sessment of drivers and passengers behaviour when facing
tion. We expect that SMA actuators will have their place an imminent side crash, 2005 IRCOBI Conference Proceed-
in automotive applications. The corresponding test methods ings, Prague, Czech Republic, 2005.
[6] S. Ruhl, M. Grinberg, and D. Willersinn, Empirical evalu-
are going to be used by car manufacturers and consumer or- ation of motion models for a side-looking driver assistance
ganisations like EuroNCAP. Last but not least, other funded system, Proceedings of the 5th International Workshop on
projects will build on the gained knowledge. Intelligent Transportation (WIT), Hamburg, Germany, 2008.
Using pre-crash sensing information, the conventional [7] J. Tandler, C. Preis, D. Willersinn, and M. Grinberg, Side pre-
passive safety devices can be signicantly enhanced. Know- crash system specication, Intelligent Vehicles Symposium,
Tokyo, Japan, 2006.
ing what is impacting when and where offers a potential [8] J. Tandler, D. Willersinn, and M. Grinberg, Data analysis
to protect the car occupant much more effectively. After of the Aprosys side pre-crash sensing system, Intelligent
frontal pre-crash sensing systems have an increasing mar- Vehicles Symposium, Istanbul, Turkey, 2007.
ket share, side pre-crash sensing systems will be offered in [9] E. Zimmerman and J. Dias, Pre-crash actuators for side im-
the near future. pact, Presentation, Haus der Technik Essen, Germany, 2006.
[10] E. Zimmerman and V. Muntean, Generic FE-simulation
model for the layout of pre-crash actuators in side impact,
Transfac Conference, San Sebastian, Spain, 2006.
Acknowledgements [11] E. Zimmerman and V. Muntean, Pre-crash actuator to im-
Several companies and research institutes worked together prove car structural performance in side impact, EAEVC
to collect the knowledge necessary to develop new tech- Conference Budapest, Hungary, 2007.