Speaker: Dr Anke Schwarze, Technische Universität Braunschweig, Department of Research Methods and Biopsychology, Germany Talk as part of the seminar series at HUman Centred Design Institute, London.

1. Theoretically guided HMI development for a driver
assistance system
Anke Schwarze
TU Braunschweig | Department of Research Methods and Biopsychology

2. Overview
 Principles of HMI-development
 classically designed HMIs
 behaviourally designed HMIs
 VIDE
 principle
 scenarios
 Assessment of VIDE
 subjective and objective data
 Adaptation to a specific situation
 Summary and outlook
May 25th 2011 | Anke Schwarze | page 2

3. Classically designed HMIs
 presentation of information about the status of the vehicle
 information has to be perceived, processed, and an adequate reaction has to
be identified and executed
 arbitrary stimulus-reaction-configurations are used
 the driver has to interpret the signal to produce the correct reaction
 the information processing of the driver is manipulated
the use of many cognitive resources is necessary
http://www.bmw.com/at/de/insights/technology/technology_guide/articles/_shared/img/info_display.jpg
http://www.maxxcount.de/maxxcount/CustomUpload/374O357O 340O370O356O 369O350O332O 352O375O375O322O366O372O365O 371O /Con_M _GATEWAY-CAN_FIS-Display2_gross.jpg
http://www.bmw.com/at/de/insights/technology/technology_guide/articles/_shared/img/multi_information_display.jpg
May 25th 2011 | Anke Schwarze | page 3

4. Reducing the amount of information processing
 Reduction of the necessary information processing of the driver if…
…behaviourally relevant information is presented
…pre-programmed and consolidated stimulus reaction
configurations are used
…the behaviour of the driver is manipulated
May 25th 2011 | Anke Schwarze | page 4

5. Classical and behavioural approach
Classical approach Behavioural approach
 information about the status of the car  behaviourally relevant signals
 use of arbitrary stimulus-reaction  use of pre-programmed and
configurations consolidated stimulus-reaction links
 manipulation of the information  manipulation of the behaviour of the
processing of the driver driver
communication coupling of behaviour
May 25th 2011 | Anke Schwarze | page 5

6. HMI development
Our „rational“ approach levels of behavioural
control (Eggert, 2007)
 general strategy of developing HMIs
insight
complexity
 no substitution of cognitive systems
instructions
 analysis of the situation, identification of the models
optimal reaction
operant conditioning
 finding out how this reaction can be provoked classical conditioning
levels of behavioural control
imprinting
 if a fast and simple reaction is necessary:
low level of behavioural control instincts
stability
 if a complex reaction is necessary and there
reflexes
is no time pressure: high level of behavioural
control
May 25th 2011 | Anke Schwarze | page 6

7. Three core questions in the design of ADAS
 What does the behavioural repertoire of the driver consist of?
 Which situation-specific reaction should be elicited?
 How can the desired reaction be elicited most efficiently?
May 25th 2011 | Anke Schwarze | page 7

8. APPLYING THE APPROACH IN THE
DESIGN OF A DRIVING
ASSISTANCE SYSTEM
May 25th 2011 | Anke Schwarze | page 8

9. The assistance system – characteristics
 the assistance system should assist the driver during his normal driving
task
 the HMI should be able (as a vision) to replace the classical cockpit
instrumentation (e.g. tachometer…)
 all relevant information should be presented as easy as possible
 pratical implications (financial, legal, technical limitations…) are of
secondary importance in the first concept stage
May 25th 2011 | Anke Schwarze | page 9

10. Design of the HMI – main principles
 behaviourally relevant information instead of status information
 no information about the actual driving speed but information about the
appropriateness of the driving speed
 no information about the absolute distance to another car but information
about how dangerous the situation is („risk potential“)
 based on the behavioural repertoire of the driver
 longitudinal control: accelerating, braking
 lateral control: steering
 Which stimuli control behaviour?
May 25th 2011 | Anke Schwarze | page 10

11. Spatial orientation
 spatially oriented reactions in driving (to the left, to the right,
forward, backward)
 it seems reasonable to present stimuli spatially oriented as
spatially oriented stimuli gain easily control over spatially oriented
reactions
May 25th 2011 | Anke Schwarze | page 11

12. Stimuli
 use of the dashboard behind the steering wheel
 visual stimuli
May 25th 2011 | Anke Schwarze | page 12

13. Dynamic task – dynamic visualization
 dynamic task
 the behaviour of the driver changes the illustration in the display in a
compatible way
accelerating
appropriate
decelerate
too fast
too slow
May 25th 2011 | Anke Schwarze | page 13

14. Appropriateness of driving speed
driving speed is appropriate
regarding the traffic situation
car symbol is in the middle of the
display
driving speed is inappropriate
(too high) regarding the traffic
situation
car symbol is shifted to the top of
the display
May 25th 2011 | Anke Schwarze | page 14

15. Appropriateness of the distance to an object ahead
 if the distance to an object ahead
is too small a risk potential arises
 this is illustrated by a longitudinal
colour gradient
May 25th 2011 | Anke Schwarze | page 15

16. Appropriateness of the distance to the front car
distance to the car ahead is too
small
yellow colour gradient from the
top
distance to the car ahead is critical
red colour gradient from the top
May 25th 2011 | Anke Schwarze | page 16

17. Appropriateness of the lateral position
distance to the left lane marking is
too small
yellow colour gradient from the
left
May 25th 2011 | Anke Schwarze | page 17

18. Appropriateness of the lateral position
distance to the left lane marking is
too small/critical
yellow/red colour gradient from
the left
distance to the right lane marking is
too small/critical
yellow/red colour gradient from
the right
May 25th 2011 | Anke Schwarze | page 18

19. Further principles and elements
 safety signal: car symbol is located in the middle of the
display, no colour gradients are visible
presentation of signals only if the driver has to change his
behaviour (shifting the car symbol, colour gradients)
 additional important information in specific critical situations
 if particular functions of the vehicle achieve a critical status,
e. g. the level of oil is far too low, VIDE gives short
commands like
“Please pull over to the side of the road and stop the
engine.”
May 25th 2011 | Anke Schwarze | page 19

20. Assessment of VIDE (Goralzik, 2009)
Hypotheses
Driving with VIDE results in smaller integrals of speeding for each characteristic of
the track than driving with the classical instrumentation.
This should hold for different courses of the track and zones of different speed
limits.
May 25th 2011 | Anke Schwarze | page 20

21. Assessment of VIDE – Method
Configuration stage realised in the study
 appropriateness of driving speed regarding the
 given speed limit
 course of the track
 appropriateness of the
 distance to the lane markings
May 25th 2011 | Anke Schwarze | page 21

22. Assessment of VIDE – Method
 HMI-lab of the German Aerospace
Center in Braunschweig
 26 participants (mdn=38 years,
range=24-59 years)
 at least five years of driving practice
(mdn=22 years, range=7-42 years)
May 25th 2011 | Anke Schwarze | page 22

23. Assessment of VIDE – Method
km/h- course of the track
max.
rural road motor-
way
straight left-hand right-hand
bend bend
60 + + +
80 + + + +
100 + +
120 +
130 +
May 25th 2011 | Anke Schwarze | page 23

24. Assessment of VIDE – Method
 comparison of driving with VIDE and
driving with classical instrumentation
(tachometer, revolution counter)
 no vehicles ahead
classical
instrumentation
VIDE
May 25th 2011 | Anke Schwarze | page 24

25. Assessment of VIDE – Method
 questionnaire to investigate several aspects of acceptance
 subjective perception of safety
 comprehensibility
 comfort
 usefulness
 perceived control
 perceived degree of efficiency
 irritability
 conformity with expectancies
 appropriateness of driving behaviour
 intentions to use the system
 requests for modifications concerning the design of VIDE
May 25th 2011 | Anke Schwarze | page 26

26. Assessment of VIDE – Analysis
 calculation of the integral of the actual driving speed [m/s] as a function of
simulator runtime [s] when the given speed limit had been exceeded
 larger integrals imply a larger extent of speeding
 analysis of the difference between the averaged integral while driving with VIDE
and while driving with classical instrumentation with a permutation test for
dependent samples (statistic: difference of means)
May 25th 2011 | Anke Schwarze | page 27

27. Assessment of VIDE – Significant results
straight parts
 80 km/h: t(25)=-5,36; p<0,01
 100 km/h: t(25)=-2,74; p<0,01
left-hand bends
 80 km/h: t(25)=-3,64; p<0,01
right-hand bends
 80 km/h: t(25)=-3,98; p<0,01
 driving with VIDE caused less speeding in some of the configurations of the
track when compared to the classical instrumentation
May 25th 2011 | Anke Schwarze | page 28

28. Assessment of VIDE – Questionnaire
Preferences of VIDE in comparison to the classical instrumentation
Item %
„Classical „New
display“ display“a
The information of which display made you feel
30.8 69.2
safer?
Which display do you deem more reasonable? 38.5 61.5
Which display influenced you more while driving? 23.1 76.9
Which display made you drive more
11.5 84.6
appropriately?b
„No“ „Yes“
Do you believe that the new displaya allows for
safer driving? 7.7 92.3
Note. aThe notation ”New display“ corresponds to VIDE. bMissing values cause the categories not to sum up
to 100 per cent.
May 25th 2011 | Anke Schwarze | page 29

29. Research project „FAMOS – Galileo for Future Automotive Systems“
(funded by the Federal Ministry of Economics and Technology)
ADAPTATION OF VIDE TO A
NEW SITUATION
May 25th 2011 | Anke Schwarze | page 30

30. Merging into motorway traffic
the driver is assisted in
 changing from the acceleration
lane onto the motorway
 adapting driving speed
 keeping an appropriate distance
 lane keeping
no automatic lane change!
DLR
development of different levels of assistance (German Aerospace Center)
 basic functions - finding a gap - guidance to a gap - partly automated
May 25th 2011 | Anke Schwarze | page 31

31. Dashboard of Volkswagen
 multi-function display between revolution counter and tachometer
 need for information about the driving speed (field tests)
original image: www.volkswagen.de
May 25th 2011 | Anke Schwarze | page 32

32. Adaptation to a lane change
the main elements of VIDE are kept
 appropriateness of the distance to a vehicle ahead and to the lane
markings illustrated by colour gradients
 appropriateness of driving speed illustrated by the dynamic shift of the
car symbol
the new element
 vehicles at the motorway produce a risk potential during and after a
lane change
 it is projected how large the distance to the other vehicles will be during
and after a lane change, in dependence of the driving speeds
 anticipated risk potential if a lane change is performed right now
illustrated by an additional lateral colour gradient
May 25th 2011 | Anke Schwarze | page 33

33. Design
 transfer of the principles into a
perspective design
 all elements remain
 perspectivic adaptation
 realistic car symbol
 fitting the information of the level
of assistance and the mode into
the display
May 25th 2011 | Anke Schwarze | page 34

34. Acoustic signals
additional assistance by acoustic signals in specific situations
 e.g. lane change is possible, level of assistance is changed, strategy is
changed
 guiding attention
 enhancing attention
 example: „gong“ sound which signals the possibility of a lange change
without a risk potential
May 25th 2011 | Anke Schwarze | page 35

35. Summary
 integrated HMI concept for different assistance functions
 integration of behaviourally relevant data from different sources
 concept of a universal interface between car and driver
 possibility of coupling the behaviour of the driver and the car via an adequate
HMI
 possibility of integrating additional information in specific situations
May 25th 2011 | Anke Schwarze | page 36

36. Outlook
 integrating more information in order to calculate the appropriateness of
driving manner more accurately
 adapting VIDE to the individual driver in order to effectively influence his
behaviour towards a safer driving manner
 identification of further situations for which the display has to be
expanded or modified
May 25th 2011 | Anke Schwarze | page 37

37. Thank you very much for your attention!
Dr. Anke Schwarze
anke.schwarze@tu-bs.de
Technische Universität Braunschweig
Department of Research Methods and
Biopsychology (IPMB)
May 25th 2011 | Anke Schwarze | page 38