Andrei Aksjonov

DRIVER DISTRACTION WHILE PERFORMING A SECONDARY TASK EVALUATION APPROACH BASED ON FUZZY EXPERT SYSTEM Smartphone usage as a driver’s secondary activity: Case study Andrei Aksjonov1, Pavel Nedoma1, Valery Vodovozov2, Eduard Petlenkov2, Martin Herrmann3 1 ŠKODA AUTO

a.s., Mladá Boleslav, Czech Republic.

2 School

of Engineering, Tallinn University of Technology, Tallinn, Estonia.

3 IPG

Automotive GmbH, Karlsruhe, Germany.

This project has received funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 675999.

Content 1. ITEAM Project 2. Driver Distraction

3. HMI 4. Fuzzy Logic Evaluator 5. Case Study 6. Results 7. Conclusions and Discussions

8. Future Research

2

10th Graz Symposium VIRTUAL VEHICLE 27.06.2017


InTErdisciplinary training network in Multi-actiated ground vehicles (ITEAM): Consortium •9 European countries •5 partner organizations •11 beneficiaries

•7 nonacademic organizations •7 universities •2 research centers

https://iteam-project.net/ 3



Driver Distraction: Definition “Driver distraction occurs when a driver is delayed in the recognition of information needed to safety accomplish the driving task because some event, activity, object or person within or outside the vehicle compelled or tended to induce the driver’s shifting attention away from the driving task” – J.R. Treat.[1] “Driver distraction is the diversion of attention away from activities critical for safe driving towards a competing activity” – M.A. Regan.[2] [1] Young, K.; [2] Regan,

4

Regan, M.; Hammer, M. Driver Distraction: A Review of Literature. Distracted Driver, Sydney, NSW: Australian College of Road Safety, 2007

M.A.; Lee, J.D.; Young, K.L. Driver Distraction: Theory, Effects, and Mitigation. CRC Press Taylor & Francis Group, Boca Raton, FL, USA, pp. 31-40



Driver Distraction: Secondary and primary tasks[1] Performing maneuver

Primary Task

Reading traffic signs Monitoring traffic Overcoming weather Managing vehicle driving functions

In-vehicle features Secondary task

Managing non-driving functions Controlling comfort Entertainment Using electronic devices

Items brought into the Personal hygiene vehicle Interaction with passengers [1] Depcik,

C.; Hausmann, A.; Sprouse III, C.B. Development of an Adaptive Human-Machine-Interface to Minimize Driver Distraction and Workload. Proceedings of the ASME 2013 International Mechanical Engineering Congress & Exposition (IMECE2013), Nov. 13-21, 2013, San Diego, California, USA. 5



Driver Distraction: Modes[1]

Visual

[1] Depcik,

Auditory

Manual

Cognitive

C.; Hausmann, A.; Sprouse III, C.B. Development of an Adaptive Human-Machine-Interface to Minimize Driver Distraction and Workload. Proceedings of the ASME 2013 International Mechanical Engineering Congress & Exposition (IMECE2013), Nov. 13-21, 2013, San Diego, California, USA. 6



Driver Distraction: Evaluation methods Off-line

Questionnaire survey

Comparative driving

Note taking

NASA Task Load Index

Situation Awareness Global Assessment Technique

On-line

Psychologic

Psychomotor

Longitudinal dynamics

7

Lateral dynamics


Hearth rate

Skin electric potential

Visual

Brain waves

Head movement

Eye tracking


Human-Machine Interface (HMI)

Machine

Human

HMI

Environment

8



HMI: Vehicle control loop

Actuators Brake pedal Steering wheel Throttle pedal

Sensors

User control

Driver

Steering wheel buttons Switches around steering wheel Dash panel buttons and knobs Voice recognition Multifunctional haptic controllers User feedback Haptic Visual Audible

9


Vehicle

Processor

Wheel speed Lateral / longitudinal acceleration Yaw rate Brake pressure Steering wheel angle Sensors Camera Radar Laser GPS

Environment


HMI: Driver distraction detection and evaluation algorithm Performance

Driver

Actuation Distraction detection

Performance prediction

10


Vehicle

Dynamics

Secondary task

Environment

Road


rn(t)

Inference mechanism

Rule-base

Modus ponens form expert knowledge

11


Defuzzification

r1(t) r2(t) r...(t)

Fuzzification

Fuzzy Logic Controller (FLC): Conventional controller

y(t)

u(t) Process

Fuzzy sets ↓ numbers


FLC: Fuzzification 1 close

far

out

μdx

+

μdx

+ 0 1

𝝁𝒅𝒙 𝒐𝒖𝒕

dx dxmax good

μdv

𝑑𝑥 𝜇𝑐𝑙𝑜𝑠𝑒 𝑎 = 𝝁𝒅𝒙 𝒇𝒂𝒓

bad

awful

Outer product*

+

+ 0

12

dv 10th Graz Symposium VIRTUAL VEHICLE 27.06.2017

𝐶= 0 0

𝒅𝒗 𝝁𝒅𝒙 𝒇𝒂𝒓 ∙ 𝝁𝒃𝒂𝒅 0

𝒅𝒗 𝝁𝒅𝒙 𝒐𝒖𝒕 ∙ 𝝁𝒈𝒐𝒐𝒅 𝒅𝒗 𝝁𝒅𝒙 𝒐𝒖𝒕 ∙ 𝝁𝒃𝒂𝒅 0

𝝁𝒅𝒗 𝒈𝒐𝒐𝒅 𝑏=

μdv

𝒅𝒗 0 𝝁𝒅𝒙 ∙ 𝝁 𝒇𝒂𝒓 𝒈𝒐𝒐𝒅

dvmax

𝝁𝒅𝒗 𝒃𝒂𝒅 𝑑𝑣 𝜇𝑎𝑤𝑓𝑢𝑙

 b1   b1  a1 b1  a2 b  b  a b  a 2 T  a1 a2  an    2 1 2 2 * C  b  a  ba         bm  bm  a1 bm  a2

 b1  an   b2  an       bm  an 


FLC: Rule-base a If a is MF1 and b is MF1 Then R is r11

R

If a is MF1 and b is MF2 Then R is r21

MF1

r11

r12

…

r1n

MF2

r21

r22

…

r2n

…

rmn


… MFm

rm1

rm2

…

13

 r1n   r2 n        rmn 

…

 r11 r12 r r22 21 R    rm1 rm 2

…

b

…

MF1 MF2 … MFn


FLC: Inference mechanism and defuzzification 𝑟1 𝑅 = 𝑟1 𝑟2

𝑟2 𝑟3 𝑟4

𝑟3 𝑟4 𝑟5

𝒅𝒗 0 𝝁𝒅𝒙 𝒇𝒂𝒓 ∙ 𝝁𝒈𝒐𝒐𝒅

𝐶= 0 0

0 Hadamard product*

𝐷 =𝐶∘𝑅 = 0 0

n m

d u

i 1 j 1 n m

14

𝒅𝒗 𝝁𝒅𝒙 𝒐𝒖𝒕 ∙ 𝝁𝒃𝒂𝒅 0

𝒅𝒗 𝝁𝒅𝒙 ∙ 𝝁 𝒇𝒂𝒓 𝒈𝒐𝒐𝒅 ∙ 𝑟2

𝒅𝒗 𝝁𝒅𝒙 ∙ 𝝁 𝒐𝒖𝒕 𝒈𝒐𝒐𝒅 ∙ 𝑟3

𝒅𝒗 𝝁𝒅𝒙 ∙ 𝝁 𝒇𝒂𝒓 𝒃𝒂𝒅 ∙ 𝑟3 0

𝒅𝒗 𝝁𝒅𝒙 ∙ 𝝁 𝒐𝒖𝒕 𝒃𝒂𝒅 ∙ 𝑟4 0

ij

;

i  1,2,..., n;

c i 1 j 1

𝒅𝒗 𝝁𝒅𝒙 𝒇𝒂𝒓 ∙ 𝝁𝒃𝒂𝒅 0

𝒅𝒗 𝝁𝒅𝒙 𝒐𝒖𝒕 ∙ 𝝁𝒈𝒐𝒐𝒅

ij


j  1,2,..., m

 c11  r11 c12  r12 c r c22  r22 * D  C  R   21 21     cm1  rm1 cm 2  rm 2

 c1n  r1n   c2 n  r2 n       cmn  rmn 


FLC: Outlook dx = [0 1.5] close far out

DD dv = [0 12]

15

good

0

14.3

42.9

bad

0

57.2

85.8

awful

28.6

71.5

100

Parameter

Fuzzy evaluator

Structure

MISO

Crisp input

dx (3 MFs) dv (3 MFs)

Crisp output

DD (8 MFs)

MFs

Linear Symmetric

Inference method

Matrix (Sugeno´s)

Rule-base

9 Modus Ponens

Defuzzification

Geometric center



Case Study: Participants Age є [24 39] 18 participants

• Mean 30

38.9% ≥ 30

27.8% female 72.2% male

Experience є [1 21] • Mean 11

50.0% ≥ 11 61.1% < 30

50.0% < 11

Special thanks to the volunteers from IPG Automotive GmbH (Karlsruhe, Germany) for participating in the driver-in-the-loop experiments. 16



Case Study: Apparatus • • • •

Gas pedal Break pedal Steering wheel 2 LCDs

• Real-time driver-inthe-loop simulation • Head-up display • MATLAB®/Simulink® integration with IPG CarMaker® System Experience Platform driving simulator

17



Case Study: Task and procedure

90

Length 10626 m 2 lines Line width 3.5 m

30

50 • Sample frequency 50Hz (0.02 s sample period) 18



Results: Driver performance dv [km/h]

dx [m]

19

ok

ok

t [cs]

t [cs]



Results: Fuzzy logic evaluation

20



Results: Resultative level of distraction calculation DD [%]

y1

y2

t [s]

y

t2-t1 y1 y2

(𝑡2 − 𝑡1 ) 𝐴𝑖 = (𝑦1 + 𝑦2 ) ∙ 2

Ai

𝒏

𝑨𝒏 = ෍ 𝑨𝒊 t1 t2 21


t

𝒊=𝟏

𝑫𝑫𝒍𝒆𝒗𝒆𝒍 =

𝑨𝒏

𝒕𝒕𝒐𝒕𝒂𝒍

[%]


Results: Resultative level of distraction surface

𝑫𝑫𝒍𝒆𝒗𝒆𝒍 = 𝟐𝟖. 𝟓𝟔 %

22



Results: Driver-in-the-loop experiment outlook Gender 50.00

Age

Experience

46.65 42.11 38.49

40.00

DDlevel [%]

32.29

30.00 20.00 10.00 0.00

23


34.87

30.45


Conclusion and Discussions ✓ Evaluation mechanism development based on Fuzzy Set theory; ✓ Real-time driver-in-the-loop experimentation; ✓ Smartphone usage while driving case study.

o Male driver handles the secondary task distraction better than female drivers; o Drivers younger than 30 years old performed the secondary task better than older ones; o Drivers with driving experience longer than 11 years deal with the distraction situation better than novice drivers.

24



Future Research: Driver performance prediction model DD [%]

Secondary task

Secondary task

dx t [s]

Vl r

dx p  dx dxr 0, dxr   dx  dx p , dx p  dx

dxp Trained Network

dvp

dv p  dv dvr 0, dvr   dv  dv p , dv p  dv

dv 25


DD


Questions

? ?? ? ? ? ??

? ? ? ?? ? ??? ? ?? ? ?

?? ? ? ? ? ? ? ? 26


THANK YOU FOR YOUR ATTENTION

CONTACT INFO: Andrei Aksjonov, MSc ŠKODA AUTO a.s., Tř. Václava Klementa 869, 293 01 Mladá Boleslav II, Czech Republic +49 157 8899 8333, [email protected], www.skoda-auto.com