Drowsiness problem The estimation of Drowsiness is an - - PDF document

The estimation of drowsiness under different conditions based on behaviour observations from a simulator study Drowsiness problem

• Drowsiness is an intermediate condition between alertness and

sleep which reduces the consciousness level and hinder a person to respond quickly (Awais, Badruddin, & Drieberg, 2017)

• Drowsy driving has caused about 2.5 percent of fatal accidents

from 2011 through 2015 in the USA (Correa, Orosco, & Laciar, 2014)

• Contributed to 22%-24% of the crashes or near-crashes risks

(Klauer, Dingus, Neale, Sudweeks, & Ramsey, 2006)

• German Road Council (DVR) - one out of four fatal highway

crashes has been caused by drowsy drivers (Husar, 2010)

WACHsens project

• Partners

– Human Research Institut für Gesundheitstechnologie und Präventionsforschung GmbH – Institut für Fahrzeugtechnik, Technische Universität Graz – AVL United Kingdom

• Aim

– To develop a method in order to warn about impaired fitness to drive with the primary goal of drowsiness detection

• Objectives:

– to collect and merge data from different sensor as well as camera monitoring and driver behaviour observations gathered from simulator test drives – to identify changes in the attention, especially when driving automated, and to identify possible safety-relevant effects. – Observation: to estimated the tiredness level to combine it with the other data and to see if there are differences under different driving conditions

Study procedure

• 100 test persons gender and age balanced
• 2 states, well-rested and tired

– tired state: awake for 16 h continuously or the previous night with half of the sleeping time.

• 2 different driving modes in the simulator: manual and

automated (LK and CC) – manual-rested, automated-rested and manual-tired, automated-tired

• 30 minute monotonous drive on a highway during twilight

without traffic events

• avoid drinking alcohol or an unusual amount of caffeine-

content drinks

• Data collection
• vehicle-based measures:

– speed – steering wheel movement – position

• physiological signals:

– EEG and ECG – skin conductivity and respiration – eyelid and eye movements

• four cameras position

– front – face – 2x from the side

Observation procedure

• Two observers
• Using observation tool (excel based) for data

input and first calculations

• pausing video at each new activity of the

testperson

Observation variables I

• Position and activities of

– both arms (position on the steering wheel or other positions, scratching faces etc.) – both hands (short movements of the hands, movements

• f the fingers)

– the upper body (short movements, leaning forward, backwards, to the side, sighing etc.) – the head (short movements, looking to the side, leaning forward, backwards, to the side etc.)

• Activities of the face (yawning, lip and tongue

movements, mimic etc.)

• Signs of sleepiness

Observation variables II

• Based on Observer Rating of Drowsiness (ORD -

Wierwille & Ellsworth, 1994) and Observer Rated Sleepiness (ORS – Anund et al. 2013)

• Measured on a 4 part scale:

– 1 = no signs of sleepiness – 2 = signs of sleepiness – 3 = shortly before falling asleep – 4 = sleeping

Sample description

Gender count % men 25 50 women 25 50 total 50 100 Age count % under 40 years 19 38 between 40 and 60 years 15 30 60 years and older 16 32 total 50 100 Driving experience count % 0-10k 4 8 10-50k 9 18 50-100k 7 14

• ver 100k

30 60 total 50 100 Experiences with driving assistant systems count % no 19 38 yes 31 62 total 50 100

Results

Observer rating /test person rating

count avarage of KSS score No or hardly signs of sleepiness 99 5,23 Signs of sleepiness but no sleep 46 6,43 Sleeping 51 8,06 total 196 6,25

• Test persons rated their tiredness level on a 9 point scale

(KSS - Karolinska Sleep Scale )

• Comparison with the observer rating (3 point scale) after

the drive

• Result is significant (the higher the observer rating the

higher the test person rating)

Observer rating / test drives

awake / automated awake / manual tired / automated tired / manual No or hardly signs of sleepiness 34,2 73,5 26,0 54,0 Signs of sleepiness but no sleep 42,1 20,4 20,0 20,0 Sleeping 23,7 6,1 54,0 26,0 total 100 100 100 100 awake tired automated manual No or hardly signs of sleepiness 56,3 40,0 29,5 63,6 Signs of sleepiness but no sleep 29,9 20,0 29,5 20,2 Sleeping 13,8 40,0 40,9 16,2 total 100 100 100 100

Observer rating / test drives

awake / automated awake / manual tired / automated tired / manual No or hardly signs of sleepiness 34,2 73,5 26,0 54,0 Signs of sleepiness but no sleep 42,1 20,4 20,0 20,0 Sleeping 23,7 6,1 54,0 26,0 total 100 100 100 100 awake tired automated manual No or hardly signs of sleepiness 56,3 40,0 29,5 63,6 Signs of sleepiness but no sleep 29,9 20,0 29,5 20,2 Sleeping 13,8 40,0 40,9 16,2 total 100 100 100 100

Observer rating / test drives

awake / automated awake / manual tired / automated tired / manual No or hardly signs of sleepiness 34,2 73,5 26,0 54,0 Signs of sleepiness but no sleep 42,1 20,4 20,0 20,0 Sleeping 23,7 6,1 54,0 26,0 total 100 100 100 100 awake tired automated manual No or hardly signs of sleepiness 56,3 40,0 29,5 63,6 Signs of sleepiness but no sleep 29,9 20,0 29,5 20,2 Sleeping 13,8 40,0 40,9 16,2 total 100 100 100 100

Observer rating / test drives

awake / automated awake / manual tired / automated tired / manual No or hardly signs of sleepiness 34,2 73,5 26,0 54,0 Signs of sleepiness but no sleep 42,1 20,4 20,0 20,0 Sleeping 23,7 6,1 54,0 26,0 total 100 100 100 100 awake tired automated manual No or hardly signs of sleepiness 56,3 40,0 29,5 63,6 Signs of sleepiness but no sleep 29,9 20,0 29,5 20,2 Sleeping 13,8 40,0 40,9 16,2 total 100 100 100 100

Observer rating / test drives

awake / automated awake / manual tired / automated tired / manual No or hardly signs of sleepiness 34,2 73,5 26,0 54,0 Signs of sleepiness but no sleep 42,1 20,4 20,0 20,0 Sleeping 23,7 6,1 54,0 26,0 total 100 100 100 100 awake tired automated manual No or hardly signs of sleepiness 56,3 40,0 29,5 63,6 Signs of sleepiness but no sleep 29,9 20,0 29,5 20,2 Sleeping 13,8 40,0 40,9 16,2 total 100 100 100 100

Observer rating / test drives

awake / automated awake / manual tired / automated tired / manual No or hardly signs of sleepiness 34,2 73,5 26,0 54,0 Signs of sleepiness but no sleep 42,1 20,4 20,0 20,0 Sleeping 23,7 6,1 54,0 26,0 total 100 100 100 100 awake tired automated manual No or hardly signs of sleepiness 56,3 40,0 29,5 63,6 Signs of sleepiness but no sleep 29,9 20,0 29,5 20,2 Sleeping 13,8 40,0 40,9 16,2 total 100 100 100 100

Test person characteristics

• No significant differences of body

activities and observer rating with regard to

– Gender – Age – Driving experience – Experience with driving assistant systems

Observer rating / activities

• Significant less changes of positions and activities of

– the left and right arm – the upper body – the head – the face – and in total

between the test person with no signs of sleepiness in comparison with those with signs of sleepiness or those who were sleeping

5 10 15 20 25 30 35 no or hardly signs signes of sleepiness sleeping no or hardly signs signes of sleepiness sleeping no or hardly signs signes of sleepiness sleeping no or hardly signs signes of sleepiness sleeping no or hardly signs signes of sleepiness sleeping left arm right arm upper body head face

First signs of sleepiness

• No significant differences with regard to gender,

age, driving experience and experience with DAS

time no or hardly signs 20:36 signs of sleepiness 10:36 sleeping 08:12 tired 10:36 awake 14:48 manual 14:48 automated 09:48 awake / automated 12:12 awake / manual 18:24 tired / automated 08:17 tired / manual 13:00

• Significant

differences between:

Conclusion

• Test persons are getting tired earlier

while driving in automated mode

• Driving “awake automated” seems to

have a similar effect than driving “tired”

• Tiredness leads to more body activities
• Gender, age, driving experience and

experience with DAS do not effect the sleepiness level or body activities

Limitation

• Due to the workload mainly one observer

but ratings are similar with regard to the subjective ratings of the test persons

• Simulator study
• „extreme“ condition (no traffic, no

passenger, no music etc.)

Thank you!! Questions????